Page 1
1
SALA 29, January 6-8, 2011, CIIL, Mysore, India Revised on January 14, 2011
Verbal and Nominal Reciprocals in South Asian Languages (SALs) and
Obligatory Order Reversal in the Nominal Anaphor in Dravidian 1
Martin Everaert , University of Utrecht ,[email protected]
Karumuri V. Subbarao, University of Hyderabad,
[email protected] , , [email protected]
1. The study of verbal reciprocal with special reference to Case Copying and Swapping
(Order Reversal) in nominal reciprocals is very interesting in SALS. While the Case
Copying strategy (Subbarao & Saxena 1989, Subbarao & Lalitha Murty 2000 for
Telugu , Amritavalli 2000 for Kannada, , Sarju Devi & Subbarao 2000) for Manipurii
is interesting, though rather rare, Swapping of the constituents of the bipartite
reflexives and reciprocals is unique and not found anywhere else universally. We shall
also show that long-distance binding of a complex nominal reciprocals found in
Marathi, (Wali 2000) too is unique. We shall demonstrate that Case (both
morphological and structural) plays a very important role in Swapping or Order
Reversal in Reciprocals (ORR) and that Case is intrinsically linked to Subject-Object
Antecedence in Dravidian. We shall attempt to provide an explanation for the
occurrence of ORR, mostly optionally but in the Dative Subject construction
obligatorily due to Case-theoretic reasons. Our paper thus demonstrates the pivotal
role that Case plays in Dravidian syntax.2
1.1 South Asian languages (SALs) have a nominal and a verbal reciprocal and their
occurrence is language-specific. While all Dravidian languages3 and Khasi (Mon-
Khmer ) have a nominal and a verbal reciprocal, their occurrence in Tibeto-Burman is
either language-specific or group specific (as is the case in the Kuki-Chin languages). 1 We are thankful to our language consultants Parameshwari Krishnamurthi for the Tamil data and
Hemanada Bisembli for the Kannada data .Thanks are also due to Peri Bhaskararao and G. Uma
Maheshwar Rao for illuminating discussions on the nature of anaphors in Telugu. 2 For a discussion of the intrinsic relation between Case and Agreement, see Subbarao (in press).
3 Malayalam is an exception as it only has the nominal anaphor, and no verbal anaphor is available (Asher
and Kumari 199 and Jayaseelan 2000).
Page 2
2
Interestingly, the occurrence of a verbal anaphor does not presuppose the occurrence
of a nominal anaphor or vice versa in SALS. Munda languages for example have only
the verbal anaphor.
In this paper we shall focus upon:
(i) the functions that the verbal reciprocal performs,
(ii) Case Copying in complex nominal reflexives and reciprocals in Dravidian and
Tibeto-Burman languages,
(iii) Swapping or Order Reversal (ORR) in Nominal Reciprocals in Dravidian and
the role of Case,
(iv) [-//-transitive] nature of the verbal reciprocal where the verbal anaphor functions
as a detransitivizer in Dravidian and many Tibeto-Burman languages, or
differentiation in tone is used as a strategy to reflect the [+transitive] and [-
transitive] dichotomy in nominal and verbal reciprocal constructions as in Angami
(Tenyidie). This phenomenon, though rather rare in SALs, is unique to SALS and is not
found any where else in any language to the best of our knowledge,
(v) long-distance binding as a unique phenomenon in Marathi (Indo-Aryan),
and
(vi) universals proposed with regard to the occurrence of the verbal reciprocal in the
Konstanz Universals project and counterevidence to the claims made there in.
The language families considered in this paper are: Austro-Asiatic (Mon-Khmer and
Munda), Dravidian, Indo-Aryan, and Tibeto-Burman.
1.1 Occurrence of the anaphor in SALs (i) only a nominal anaphor (reflexive or
reciprocal, or a simple pronoun), or
(ii) only a verbal anaphor ( a verbal clitic for the reflexive or reciprocal), or
(iii) both nominal and verbal.
(i) Nominal anaphor: Most of the Indo-Aryan languages and some Tibeto-Burman
languages.
(ii) Nominal and verbal anaphor: All Dravidian languages except Malayalam, and
many Tibeto-Burman languages and Khasi (Mon-Khmer)
Page 3
3
(iii) Verbal anaphor alone: Most of the Munda (Austro-Asiatic) languages have only a
verbal device except in Juang (Patnaik & Subbarao 2000) and Korku (Nagaraja 1999).4
1.2 Morphological complexity
1.2.1 Nominal Reciprocal: (i) Mostly polymorphemic in nature with a bipartite
structure (two morphologically independent parts) in most of the SALs.
(ii) Case Marking: In Telugu, Kannada, Tamil (Dravidian) and Manipuri (Tibeto-
Burman) the first part of the nominal reciprocal invariably gets structurally Case-
marked by the verb. The second part is a Case Copy of the subject as in Telugu,
Kannada, Tamil (Dravidian).5 The case of Manipuri (Tibeto-Burman) case marking in
nominal reciprocals is a mirror image of the case marking found in Dravidian . In
Manipuri it is the first part that gets the Case Copy of the antecedent and it is the second
part that gets structural case from the verb.
(iii) May be person and number sensitive to the features of the antecedent
1.2.2 Verbal Reciprocal: Universally monomorphemic. It may be homophonous with
the verbal reflexive or may not be. A language may have the verbal reciprocal and may
not have the verbal reflexive as in Tenyidie (Kevichusa 2007), Ao (Pangarsenla 2005).
The verbal reciprocal performs a variety of functions in SALs.
The nominal reciprocal
Telugu (Dravidian)- Direct Object position with a Nominative Subject as antecedent
okaḷḷa- ni okaḷḷu
one acc one (nom)
‘each other’
4 See Subbarao (in press) for a detailed discussion of anaphors (nominal and verbal) in SALS.
5 We shall however show that there is one exceptions to this in the DSC (Dative Subject construction).
Page 4
4
With a Dative Subject as antecedent:
okaḷḷa-
mῑda
okaḷḷa-
ki
one on one dat
‘at each other’
Note that okaḷḷu ‘one.pl, [+human]’ has the form okaḍu ‘one’ and they are independent
morphemes in Telugu as are such forms in all other Dravidian languages.
Hindi-Urdu (Indo-Aryan)
ek dūsre ko
one another acc/dat
‘(to) one another’
Both ek ‘one’ and dūsre ‘another’ are independent forms in Hindi-Urdu and they are
not sensitive to the feature of animacy.
Khasi (Mon-Khmer)
i- wƐi ya i- wƐi
hon- one acc hon- one
‘each other’
wƐi ‘one’ is an independent morpheme in Khasi.
Mizo (Tibeto-Burman)
an- māʔ- ni- leʔ an- māʔ- ni-
3 pl self pl and 3 pl self pl
‘each other’
Page 5
5
an-māʔ ‘3 pl-self’ literally stands for the third person plural is ‘3pl (pronoun or the
subject agreement clitic and māʔ is an emphatic particle that occurs to the right of a
noun or a pronoun and is not a free morpheme. Thus, it stands for ‘they-emphatic’ in
Mizo. However, in a complex reciprocal anaphor and a reflexive anaphor it can be
interpreted as ‘self’.
Munda languages do not have an indigenous nominal anaphor (See Subbarao in press
for details).
1.2.3 The verbal reciprocal
The verbal reciprocal is monomorphemic universally. May be a prefix (Mizo, Hmar,
Thadou, Tenyidie or Ao of the Tibeto-Burman group) or an infix (Munda) or a suffix
(Dravidian) to the verb.
Indo-Aryan: No verbal reciprocal (Subbarao, in press for details)
Dravidian: kon or koḷ;
Tibeto-Burman: Mizo and Hmar in, Thadou ki
Mon-Khmer: Khasi -ya-; Munda: Mundari, Santali, and Ho -pA-
In Ho, Mundari and Santali the verbal reciprocal –pA- is infixed in the verb, and it is
placed after the first syllable of the underived form . The vowel A in –pA-stands for
any vowel and it harmonizes with the stem vowel that precedes it. The absence of the
nominal reciprocal is indicated by ø.6
6 The verb form for ‘hit’ in Santali, for example, is dāl. Since the verbal reciprocal –pA- is infixed in the verb, and the verb is
split into two parts which we’ve glossed them as dā- as ‘hit1’ and –l as ‘hit2’.
Santali (Munda)
1. un.kini- kin ø dā- pāi- l-
they.dual- dual hit1- vrec- hit2-
kā- n- a
pst [-tr] fin
‘They hit each other.’
(Minegishi and Murmu 2001:104)
Page 6
6
Inherently reciprocal verbs: In SALs some verbs which indicate a reciprocal activity
such as talk, meet, marry, fight, kiss, hug etc. inherently carry the verbal reciprocal.
The verbal reciprocal is grammaticalized in such cases. In Mizo, for example, nei
‘have’; in-nei ‘vrec-have’ = ‘marry’. No identity requirement of the arguments is
required in such cases.
We shall briefly discuss the various functions that the verbal reciprocal performs in
SALs.
1.2.4 Functions of the Verbal Reciprocal:
(i) As a group activity marker as in Khasi and some Tibeto-Burman languages,
(ii) as a hortative marker (only in Khasi),
(iii) as a collaborative effort marker (- cem) (only in Khasi)
(iv) used in the formation of specific lexical items as in Mizo, Hmar, Thadou.
Mizo and Manipuri (TIbeto-Burman and Dravidian). For example, in Manipuri
(TIbeto-Burman), the compound imənnabə ‘friend’ carries the verbal reciprocal na
as friendship is a reciprocal activity.
Manipuri (TIbeto-Burman)
i- mәn- na- bә
my- similar-
(verb)
vrec- nozr
(v) as a valence Reducer or Decausativizer: The verbal anaphor reduces the valence
of the predicate in SALs. Thus, a transitive predicate is detransitivized in Dravidan
and Tibet-Burman languages (Amritavlli 2000 for Kannada, Subbarao & Lalitha
Murthy 2000 for Telugu, Lalitha Murthy and Subbarao 2000 for Mizo, Kevichüsa 2007
for Tenyidie (Angami)).
However, the verb containing the verbal anaphor also assigns accusative Case to its
direct object. This seemingly contradicting phenomenon can be accounted for if we
Page 7
7
hypothesize that the decausativizer is different from the verbal anaphor. Further
analysis with evidence is required in support of this hypothesis.
In the following section we discuss the Antecedency requirement in nominal
reciprocals and demonstrate that structural Case plays a crucial role as a result of Case-
theretc requirements, there is obligatory reversal in the order of occurrence of the
nominal reciprocals in Dravidian.
1.3 Antecedency requirement in nominal reciprocals:
The verbal reciprocal is subject-oriented, as in Dravidian or is subject as well as
object-oriented (as in Tibeto-Burman and Khasi) (for details see, (Subbarao, in press).7
Thus, we obtain two patterns:
Pattern A: The verbal reciprocal just as the verbal reflexive is subject-oriented in
Dravidian and hence, its occurrence is prohibited when there is a non-subject
antecedent.
Pattern B: In Tibeto-Burman and Mon-Khmer (Khasi) the verbal anaphor is non-
subject-oriented and hence, the verbal reciprocal occurs when there is a non-subject
antecedent.
Before we discuss Patterns A and B in detail, a brief discussion illustrating the nature of
Case Copying in Dravidian and Manipuri (Tibeto-Burman) is in order.
Subbarao & Saxena (1989) was the first paper that discusses this fascinating
phenomenon. Such a phenomenon is found Greek and Icelandic (Everaert 2000), but
neither of these languages has Swapping/ORR which we shall discuss below. The
following is an excerpt from Subbarao (in press).
“The complex anaphor in Telugu (DR) and Manipuri (TB) is the result of reduplication
and Case Copying. The nominative or dative case marker of the subject is copied on to
the second part of the anaphor. The dative subject construction (DSC) provides further
support to the proposal concerning Case Copying on the anaphor in Telugu. The dative
case marker of the subject in italics in sentence (116) is copied on to the second part of
7 We shall present evidence from Tibeto-Burman abd Khasi (Mon-Khmer) later in section .
Page 8
8
the bipartite structure of the anaphor in (116). Recall that neither a verbal reflexive nor
a verbal reciprocal is permitted in a DSC8.
Telugu (DR)
116. mamatai ki tana mīda tanai ki
Mamata dat self on self dat
kōpam vaccindi
anger came
‘Mamatai got angry with herselfi.’
Reciprocals in Telugu provide further evidence in support of our proposal concerning
Case Copying. The second part of the bipartite structure of the reciprocal carries the
null nominative case marker of the nominative subject in sentence (117) and the dative
case marker of the dative subject in (118).
NOMINATIVE SUBJECT
117. vāḷḷu okaḷḷa- tō okaḷḷu debba lāḍu-
they (nom) one with one (nom) fight-
kon- ṭunnā- ru
vrec- progr- m.p
‘They are fighting with each other.’
DATIVE SUBJECT
118. vāḷḷa- ki okaḷḷa- tō okaḷḷa- ki
they- dat one.gen- with one.gen- dat
sariggā paḍadu
well get along
8 The reason for such prohibition is the predicate in the DSC is [-transitive] and the verbal reflexive and
reciprocal itself functions as a detransitivizer in Dravidian languages. Hence, it leads to redundancy of
features in a sentence, which a language may attempt to avoid.
Page 9
9
‘They do not get along well with each other.’
[The quote ends here.]
Recall that we mentioned above that there are two patterns that emerge in SALs with
regard to Antecedency. We shall now discuss Patterns A and B in detail.
In all Dravidian languages a subject or a non-subject can antecede9 the reciprocal.
Pattern A: Telugu (Dravidian)
Subject as an antecedent to the reciprocal: Verbal Reciprocal permitted
1. abbāyilui ammāyilaj- ki okaḷḷa- ni okaḷḷui,*j
boy-pl.nom girls- dat one.pl- acc one.pl.nom
paricayam cēsu- konn
\i,*j-
ā- ru
introduce do- vrec- pst- pl
‘The boysi introduced each otheri to the girlsj.’
When a non-subject antecedes an anaphor, the verbal anaphor cannot occur. The
following examples are illustrative.
A. Direct Object as an antecedent to the reciprocal: Verbal Reciprocal not
permitted
The antecedent ammāyu ‘girls’ is case-marked accusative in (2).
2. abbāyi-lui, ammāyili-j ni okaḷḷa- ki okaḷḷa- ni *i,j paricayam
boy-pl girls acc one dat one- acc introduce
cēs- ā- ru/ *paricayam cēsu- konn.ā- ru
do pst pl introduce do vrec.pst- pl
‘The boysi introduced the girlsj to each otherj (the girls themselves, and not to the
boys).’
(Subbarao & Lalitha Murthy 2000:261)
B. Indirect Object as an antecedent to the reciprocal: Verbal Reciprocal not
permitted
The antecedent ammāyu ‘girls’ is case-marked dative in (3).
9 To the best of our knowledge, this fact was first mentioned in Pelletier (1994) for Telugu.
Page 10
10
3. abbāyi-lui, ammāyili-j ki okaḷḷa- ni okaḷḷa- ki *i,j paricayam
boy-pl girls dat one acc one- dat introduce
cēs- ē- ru/ *paricayam cēsu- konn.ā- ru
do pst pl introduce do vrec.pst- pl
‘The boysi introduced the girls j to each other j ( the girls themselves, and not to the
boys).’
Swapping or ORR (Order Reversal in Reciprocals): We shall now discuss the
variation that occurs in the order of the constituents of a complex nominal reciprocal.
When a polymorphemic anaphor occurs in a subcategorized position with subject as its
antecedent, the order in which the two constituents occurs is fixed: the first part of the
reciprocal anaphor occurs in the position in which structural case is assigned and the
second part occurs in the position in which a case copy of the subject is copied on to.
There is however an asymmetry with regard to this order between the subject and
object (direct as well as indirect) as an antecedent of the reciprocal. When an IO or DO
is the antecedent, the order in which the constituents occur can be reversed in Kannada,
Tamil and Telugu (Dravidian).10
Thus, (4) corresponding to (3) is grammatical in spite of the fact that the order of the
constituents in the reciprocal is reversed in (4) in Telugu.
Order Reversal in a sentence with a non-subject as antecedent: XY YX
4. abbāyi-lui ammāyili-j ki okaḷḷa- ki okaḷḷa- ni *i,j
boy-pl girls dat one dat one- acc
paricayam cēs- ā- ru
introduce do pst- pl
‘The boysi introduced the girls j to each other j.’
10
We shall show later that the order in the reciprocals is obligatorily reversed in the dative subject
construction with a set of specific predicates due to Case-theoretic reasons.
Page 11
11
Similarly, Swapping/Order Reversal in (2) with Direct Object as antecedent too is
permitted (data not provided).
However, when Subject is an antecedent, Swapping/Order Reversal is not permitted.
Thus, (5) corresponding to the grammatical sentence in (1) is ungrammatical.
Order Reversal in a sentence with Subject as antecedent- not permitted
(XY *YX)
5. *abbāyilui ammāyilaj- ki okaḷḷu- okaḷḷai,*j ki
boy-pl girls- dat one.p.nom one.pl dat
paricayam cēsu- konni,*j- ā- ru
introduce do- vrec- pst- pl
‘The boysi introduced each otheri to the girlsj.’
Note that the fist part of the reciprocal okaḷḷu ‘one.pl.’ is in the nominative case and it is
the occurrence of the nominative case-marked reciprocal that prohibits such swapping
in (5). We provide evidence in support of our assertion later.
Order Reversal in reciprocals in Telugu is permitted when there is a causative verb in
the matrix clause. (data not provided)
Conclusion 1: Tentatively, we may arrive at the conclusion that with Subject as an
antecedent, Swapping/ORR is not permitted, while non-subjects (DO and IO) as
antecedents, it is permitted.
We shall however revise Conclusion I and demonstrate later that it is Case (as
discussed in the Government and Binding and in the Minimalist approaches) on the
second part of the anaphor in the unmarked order that plays the crucial role in non-
reversal. We shall arrive at the revised conclusion based on:
(i) language-internal evidence from Telugu and,
(iii) cross-linguistic evidence from Tamil and Kannada.
Subject-non-subject asymmetry with regard to the order of the constituents of a
complex anaphor.
Page 12
12
The Dative Subject construction provides further evidence in support of our claim that
there is Subject-Non-subject asymmetry with regard to the order of the constituents of a
complex anaphor in Telugu. In the DSC, the dative subject can be an antecedent to an
anaphor (6) in SALs. The order of the constituents can be reversed in Telugu in the
DSC as (7) illustrates. The nominal anaphor occurs in a subcategorized position and
the occurrence of the verbal reciprocal is not permitted in such cases as the dative
predicate is intransitive. (cf. Subbarao in press for details).
Normal Order:
6. [pāpala- ki] i [okaḷḷa- mῑda okaḷḷa- ki] i
small girls- dat one.pl- on one.pl- dat
makkuva perigindi
affection increased
‘The affection increased between each otheri to the small girlsi.’ (literal translation). =
‘The affection/liking between the small girls has gone up.’
Swapping/ORR in the DSC – permitted
7. pāpala- ki okaḷḷa- ki okaḷḷa- mῑda
small girls- dat one.pl- dat one.pl- on
makkuva perigindi
affection increased
Note that Swapping/ORR in the DSC is permitted though Subject is the antecedent in
(6) and (7). Notice that the second part of the nominal anaphor in (6) is carrying the
dative case marker which is a Case Copy of the Dative Subject. We shall show below
that it is the presence of the dative case marker that saves (7) from the Case Filter.
1. Internal Evidence from Telugu: Let us now consider more data on nominal
reciprocals in the DSC. Recall that in the bipartite structure of the reciprocal in
Telugu, it is the first part of the reciprocal that is structurally case-marked and the
second part carries the Case Copy of the subject.
Page 13
13
With a limited set of dative predicates such as telusu ‘known’, gurtu ‘remembrance’,
iṣṭam ‘liking’ paricayam.ee ‘familiarity.emphatic’, jnaapakam ‘remembrance ‘ in
Telugu, the theme takes the nominative case marker.
8a. [*vāḷḷa- ki] mēmu telusu
they- dat we.exclusive’ known
‘They know us.’ Literally: ‘We are known to them.’
Note that the theme mēmu ‘we.exclusive’ cannot be dative or accusative case-marked
as the predicate telusu in (8a) is [-transitive].
Recall that in a polymorphemic reciprocal, the first part of the reciprocal is structurally
Case-marked. The tense marker of the predicate telusu ‘known’ hence assigns
nominative Case to the first part and the second part of the reciprocal carries a Case
Copy of the subject, namely, the dative case marker ki as in (8b). (8b) however is
ungrammatical in Telugu. The question that needs to be answered is: Why (8b) is
ungrammatical and how it can be resurrected?
With such predicates when a complex anaphor occurs in the DSC, the first part should
carry the nominative case marker which is assigned by the predicate and it is the
second part that should carry the Case Copy of the subject, which is the Dative case
marker as in (8) which is ungrammatical.
8. [*vāḷḷa- ki] [okaḷḷu- okaḷḷa- ki] telusu
they- dat one.nom one.obl- dat known
‘They know each other.’ (ungrammatical in the intended sense)
In the expected pattern as in (8b), Obligatory Reversal/Swapping takes place to make
it grammatical in view of the restriction that the first part of the reciprocal CANNOT
be in the nominative case in Telugu. (9) is the result of such Obligatory Swapping.
Obligatory Swapping - DAT-NOM order in the Reciprocal
9. [vāḷḷa- ki] [okaḷḷa- ki okaḷḷu] telusu
Page 14
14
they- dat one.obl- dat one.nom known
‘They know each other.’
Now an automatic explanation is available for the non-permissibility of reversal in
reciprocals with Subject as antecedent in Telugu in sentences such as (5) repeated her
for ease in reference.
5. *abbāyilui ammāyilaj- ki okaḷḷu- okaḷḷai,*j ki
boy-pl girls- dat one.p.nom one.pl dat
paricayam cēsu- konni,*j- ā- ru
introduce do- vrec- pst- pl
‘The boysi introduced each otheri to the girlsj.’
With Subject as the antecedent, the first part of the reciprocal gets accusative case-
marked and the second part of the reciprocal is always in the nominative case which is
a Case Copy of the subject in the Nominative Subject construction. If Swapping takes
place, it is the first part of the reciprocal that is in the nominative case that ends up in
the sequence. Such sequence is not permitted. Hence, with a nominative subject as the
antecedent, Swapping or Obligatory Reciprocal Reversal is not permitted . We shall
demonstrate that this explanation is valid for Tamil too.
To summarize the above discussion, in Telugu the first part of the reciprocal cannot be
in the nominative case, which appears to be a language-specific restriction. Swapping
is permitted in polymorphemic reciprocals with a non-subject as antecedent, namely, in
the direct and indirect object positions. However, in the Dative Subject construction
Swapping is permitted as both parts of the reciprocals carry a case marker which is not
in the nominative case except with predicates such as telusu ‘known’, gurtu ‘memory’
which we can term as cognitive predicates. With such predicates, Obligatory Swapping
takes place and it is such predicates that provide a clue as to why Swapping is not
permitted with a nominative subject as antecedent. We have shown that permissibility
Page 15
15
or non-permissibility of Swapping is purely due to Case-theoretic reasons. Thus, the
case of Telugu reciprocals demonstrates the significant role that Case plays in language.
We shall now present evidence from Tamil and Kannada in support of our claim that he
first part of the reciprocal cannot be in the nominative case and Swapping is obligatory
if the first part is permitted. We draw wvidence from the DSC in Tamil and Kannada
where Obligatory Order Reversal of the reciprocal is a must in the DSC.
2. Cross-linguistic evidence:
We first provide evidence from Tamil.
Evidence from Tamil: When a non-subject is an antecedent of the nominal anaphor,
the order of the constituents can be reversed in Tamil. We shall first consider the
Nominative Subject construction.
IO as an antecedent in Tamil: The first part of the complex anaphor oruvar-ai ‘one-
Dat’ carries the accusative structural case-marker ai and the second part carries a
Case Copy of the indirect object antecedent viruntiṉ ‘guests’.
10. pili [viruntiṉar - kaḷ ukkuj oruvar- ai oruvar- ukkuj
Bill guest pl Dat one_person acc one.Obl Dat
arīmukappaṭutt- iṉ- āṉ.
introduce past 3,sg,m
‘Bill introduced the guests to each other.’
Order Reversal with IO as antecedent as in (11) - permitted
11. pili l viruntiṉar-kaḷ arkaḷ- ukkuj oruvar- ukkuj oruvar- ai
Bill guest-pl Dat one_person Dat one.Obl acc
arīmukappaṭutt- iṉ- āṉ. introduce past 3,sg,m
‘Bill introduced the guests to each other.’
Page 16
16
In Tamil (Lehmann 1989) and Malayalam (Jayaseelan 2004: 229) with a dative-verb
such as teri ‘know’ the logical subject is accusative case-marked as in (12). In Tamil
the logical subject may also be nominative case-marked while in Malayalam such
optionality is not found and the theme has to be obligstorily accusative Case-marked.
(Sree Kumar, Lucknow University. p.c.).
Tamil
12. avarkaḷ- ukku avan- ai.t/aa avan teri.y ā tu they. Dat he.- Acc he.nom know neg 3,sg,n
(default)
‘They don’t know him.’
Such accusative marking is not permitted in many SALs including Telugu, as we have
shown above. The reason for the non-availability of such case marking is due to the
fact that all predicates in a dative subject construction are invariably [-transitive]
(Subbarao & Bhaskararao 2004). In Subbarao (in press), the accusative marker that
occurs with the theme in Tamil in (12) and Malayalam in the DSC was shown to be a
specificity marker and not a structurally assigned case marker, as the verb in the
Dative Subject construction is [-transitive] and hence, it cannot assign accusative case.
Note that in Tamil in the Dative Subject construction with a predicate such as teri
‘know’ with Dative Subject as antecedent, the second part of the reciprocal may either
carry the specificity case-marker –ai as in (13) or the nominative case marker as in
(15). The nominative case marker in Dravidian is null.
DAT –ACC Order in the Reciprocal
13. avarkaḷ - ukkui oruvar- ukkui oruvar- ai.t teri.y ā tu
they. Dat one_person Dat one.Obl Acc known neg 3,sg,n (default)
‘They don’t know each other.’ DAT-ACC Order of the reciprocal-
permitted
Page 17
17
Order Reversal with Dative Subject as antecedent Reversal of the reciprocal –
permitted (ACC-DAT Order of the reciprocal) in (12)) -permitted (XY YX)
14. avarkaḷ- ukkui oruvar- ai oruvar- ukkui teri.y ā tu they. Dat one.Obl Acc one.Obl Dat known neg 3,sg,n
(default) ‘They don’t know each other.’
Tamil also permits the NOM-DAT pattern in the DSC.
DAT-NOM Order in the Reciprocal
With the Dative Subject as antecedent –Unmarked Order of the reciprocal –permitted
15. avarkaḷ - ukkui oruvar- ukkui oruvar teri.y- ā- tu
they. Dat one_person Dat one.nom known- n-eg 3,sg,n (default)
‘They don’t know each other.’
Just as in Telugu, Order Reversal in Tamil too results in ungrammaticality.
Order Reversal with Dative Subject as antecedent –Reversal of the reciprocal –not
permitted (NOM-DAT Marked Order of the reciprocal in (16)) XY *YX
16. *avarkaḷ- ukkui oruvar- oruvar- ukkui teri.y ā tu they. Dat one.Obl one.Obl Dat known neg 3,sg,n (default)
‘They don’t know each other.’
The reason for the ungrammaticality of (16) in contrast to the grammaticality of (15) is
due to the fact that a nominative case-marked anaphor in the first position is not
permitted in Tamil too, just as in Telugu.
Page 18
18
Thus, Tamil too provides evidence in support of our claim that the first part of the
reciprocal cannot be in the nominative case. The accusative case marker functioning as
a specificity marker provides a ‘Case Shield’ and it protects the sentence from case
filter after Order Reversal in sentences such as (12).
Kannada too permits Swapping the details of will be added soon.
In Malayalam there is no verbal reflexive or reciprocal and the nominal reciprocal used
for antecedents which are [+human] is monomorphemic in contrast to the
polymorphemic anaphor which is exclusively used for [-humanas]. This category
includes the categories of animals and inanimate objects (Sreekumar, p.c.).
The verbal reciprocal and the the verbal reflexive function as detransitivizers in SALs
(see Subbarao in press for details). In the following section we provide an interesting
case where tone is used as a mechanism for anti-causativization or detransitivization.
1.4 The Verbal Anaphor as [-transitive]
The Role of tone and the verbal reciprocal in Angami (aka, Tenyidie) (Tibeto-
Burman)
The verbal reciprocal ke occurs to the left of the verb in Angami.
Angami (TB)
(16a) uni-e i e lešә thu
they (dual) nom letters write
kei - tsə- ya
vrec- give- pres
‘They (dual)i write letters to each otheri.’
(Kevichusa 1999)
Angami presents a very fascinating case with regard to the choice of tone on the verb in
reciprocal constructions. The tonal variation depends upon the choice of the verbal or
nominal anaphor and as we shall show below, both the verbal and nominal anaphor
cannot occur in the same sentences as it leads to a tonal conflict Kevichüsa (2007).
Page 19
19
(i) Angami has only a nominal reflexive, and it has no verbal reflexive.
(ii) It has a nominal reciprocal huo-nie-huo ‘some-dual-some’ (‘each other’),
which is a reduplicated form.
Nominal reciprocal: When the nominal reciprocal occurs, a monotransitive verb has
falling–rising tone, and such verbs permit pro-drop. Kevichüsa (2007:107) labels them
as VA verbs.
Angami (TB)
16 b. unie- e thisɔnhie huo-nie-
huo
ŋǔ- ya
they (dual) erg every day each
other
see- pres
‘They see each other everyday.’
(Kevichüsa 2007:105)
When the verbal reciprocal ke occurs, the verb has rising - falling tone. and such verbs
with a rising - falling tone do not permit pro-drop. Kevichüsa (2007: ibid) labels them
as VO verbs.
17. unie- e thisɔnhie ke- ŋû- ya
they (dual) erg every day vrec see pres
‘They see each other everyday.’
17. unie- e thisɔnhie ke- ŋû- ya
they
(dual)
erg every
day
vrec see pres
‘They see each other everyday.’
(Kevichüsa 2007:105)
(iii) The nominal reciprocal and the verbal reciprocal cannot cooccur, for there is a
Tone Clash/Tonal Conflict.
Page 20
20
18. *unie- e thisɔnhie huo-nie-
huo
ke- ŋû- ya/ ŋǔ- ya
they (dual) erg every day each other vrec see pres see pres
‘They see each other everyday.’
(Kevichüsa 2007:105)
The nominal reciprocal requires that the tone be falling–rising, and the verbal
reflexive and reciprocal requires a rising - falling tone. Both requirements cannot be
met.
Kevichüsa (2007:107) observes: ‘The cooccurrence of the nominal and verbal
reciprocal results in ungrammaticality as illustrated in sentence (41) [sentence (18)
above.]. The fact that the nominal reciprocal does not occur with the verbal reciprocal
indicates that the verbal reciprocal detransitivizes the verb thereby decreasing the
valence of the verb.’
The phenomenon of using tone for detransitivization is unique and to the best of our
knowledge we have not found it any where in the existing literature.
We shall now mention a claim made in Reinhart & Siloni (2005) with regard to the [-
transitive] nature of the verbal anaphor. While there is ample evidence from SALs in
support of their contention as demonstrated in Subbarao (in press), there a major issue
with regard to the assignment/checking of Case of the nominal anaphor when a verbal
anaphor occurs.
Reinhart & Siloni (2005:404) argue that “The selection of se [in French] reduces the
verb’s ability to check accusative Case (30a)” [(i) below from French]. The letter ti in
(i) stands for trace.
French
19. Jeani sei lave ti
John SE washes
‘Jean washes himself.’
Page 21
21
In many Tibeto-Burman and Dravidian languages when the verbal anaphor occurs. the
DO needs to have its accusative Case assigned/checked and hence, their claim cannot
be accepted. We provide a brief account of the cases Mizo (Tibeto-Burman) and
Telugu (Dravidian).
In Mizo, the nominal anaphor is a complex form a-maʔ leʔ a-maʔ ‘(s)he-self and (s)he-
self’. It is formed by conjoining two nominal forms of a-maʔ ‘(s)he-self’ with the
conjunction leʔ. The VR -in- occurs to the right of the verb stem. Just as in Dravidian
languages, the occurrence of the VR is obligatory when an anaphor occurs in a
subcategorized position (Lalitha Murthy and Subbarao 2000).
When the anaphor occurs in direct object position, the VR detransitivizes the predicate
as a result of which ‘the ergative marker [which occurs obligatorily with transitive
verbs – KVS] does not occur with the subject…’ (Lalitha Murthy and Subbarao
2000:785). That is, the subject is not in ergative case even in a transitive sentence, when
the DO is an anaphor.
Mizo (TB)
Anaphor in direct object position
20. zovai (a- maʔ leʔ a- maʔ) a- in- huā
Zova he self and he- self 3s- vr- hate
‘Zovai hates himslfi.’
(Lalitha Murthy and Subbarao 2000: 790)
We quote a discussion from Subbarao (in press). Subbarao (ibid) observes: “Note that
in (10) [ (20) above)] the nominal reflexive a-maʔ- leʔ-amaʔ that occurs in the direct
object position is not morphologically accusative case-marked. However, it is the
occurrence of the reduplicated form that does not necessitate overt morphological case
marking. It is structurally-case marked.”
However, when the monomorphemic form of the reflexive occurs, overt morphological
Page 22
22
accusative case-marking is required as in (21) below.
Subbarao (ibid) observes: ‘Such accusative case marking either overtly in (11) [ (21)
below)] or structurally in (10) [ ( 20 ) above)] creates a paradoxical situation where the
ergative marker on one hand does not occur with the subject as the predicate with the
verbal reflexive is [–transitive], and on the other hand, the [–transitive] predicate needs
to assign accusative case to the nominal anaphor, a direct object. Such anomalous
situation poses a problem for any theory to provide an explanation.
11. lālii (a-maʔi- cu) a- in- that
Lali she-self- acc 3s- vr- kill
‘Lalii killed herselfi.’
(Lalitha Murthy and Subbarao 2000: 790)
In Telugu, Tamil and Kannada (Dravidian) reciprocal constructions too such a situation
arises. In Telugu the first part of the polymorphemic anaphor is structurally case-
marked by the accusative Case-marker –ni in spite of the occurrence of the verbal
reflexive in (22) and of the verbal reciprocal in (23) below.
Verb as an accusative Case-assigner in spite of the occurrence of the Verbal reflexive
Telugu (DR)
22. rādha (tana- ni (tanu)) poguḍu-
Radha self- Acc self (nom) praise-
kon- di
vr- agr
‘Radha praised herself.’
(Subbarao and Lalitha Murthy 2000: 233)
Verbal reciprocal as an accusative Case-assigner
Page 23
23
23. abbāyilui okaḷḷa- ni okaḷḷui,*j poguḍu-
boy-pl.nom one.pl- acc one.pl.nom praise-
konni- ā- ru
vrec- pst- pl
‘The boysi praised each otheri’
Note that the first part of the complex anaphor okaḷḷu ‘one.pl’ is morphologically
accusative Case-marked by -ni, the accusative case marker in Telugu. A similar
situation arises in Tamil and Kannada too.
The discussion above clearly demonstrates that while an anaphor in a subcategorized
position in Telugu and the other Dravidian languages (except Malayalam) requires the
occurrence of the verbal anaphor, the nominal anaphor in that position requires to be
accusative Case-marked structurally. If one wishes to argue that the verbal anaphor that
occurs as a verbal anaphor is different in its transitive nature though homophonous with
the anti-causative as in (24) and ( 25) below in Mizo and Telugu respectively, the
prohibition of the ergative marker in Mizo creates a problem. Hence, this issues further
research.
Mizo (Tibeto-Burman)
[-transive]
(24)
in- hon
inchoative open
‘be opened’
[+transive]
Page 24
24
hon
open
‘open’
(25)
Telugu (Dravidian)
[-transive]
terucu- kon
open- decausativizer
‘be opened’
[+transive]
teru
open
‘open’
1.5 Long-distance Binding of the polymorphemic anaphor in Marathi (Indo-
Aryan)
Marathi is the only South Asian language that we knoe of that permits long-distance
binding of a polymorphemic anaphor. To the best of our knowledge, any such case in
not re[prted from any other language so far.
We provide an excerpt from Subbarao (in press):
“Marathi (IA) permits long-distance binding of a polymorphemic anaphor. The anaphor
(āplyā) swatāh-lā ‘self self-acc’ occurs in the direct object position in sentence (76)
[(26) below], and it can be coindexed either with lili ‘Lili’, and šām ‘Sham,’ the
subjects of the matrix and embedded clauses respectively.”
Marathi (IA)
Page 25
25
26. [S1lilii- ni šāmj- lā [S2PROj (āplyā)
Lili- erg Sham- dat ø self
swatāh- lā i,j bockār-u S2] dila S1]
self- acc pinch-inf allowed
‘Lilii allowed Shamj to pinch self i, j’ (Wali 2000: 534)
No study has thus far provided an explanation to account for the long-distance binding
of the polymorphemic anaphor in sentence (76).”
1.6 Universal and counter-evidence from SALS: SALS do not conform to some of
the universals proposed for verbal reciprocals and pose interesting counterexamples.
From Konstanz Universals Archive, we quote:
NO 15: IF basic order is OV, THEN verbal modifiers like those for negation,
causation, and reflexive or reciprocal are placed after verb roots.
In Kuki-Chin languages and in Angami and Ao the verbal reciprocal occurs to
the left of the verb.
NO 1190: IF non-subject can control reflexives/reciprocals, THEN subjects can,
too.
A dative subject cannot ‘control’ an anaphor in the DSC unless Obligatory
Swapping takes place in Telugu and Tamil.
NO.2014 There are no (clear cases of) long-distance reciprocals.
Marathi is a counter-example (Wali 2000).
1.7 Conclusion
To be written
Abbreviations:
agr: agreement, acc: accusative, cem: collaborative effort marker, dat: dative, DO:
direct object, erg: ergative, fin: finte, hon: honorific, IO: indirect object, nom:
nominative, progr: progressive, pst: past, sg: singular, pl/p: plural, tr: transitive, vr:
verbal reflexive, vrec: verbal reciprocal
Page 26
26
References
2000 Everaert, Martin. ‘Types of Anaphoric Expressions: Reflexives and Reciprocals’,
In Z. Frajzyngier & T. Curl (eds.) Reciprocals. Forms and Functions, Typological
Studies in Language, John Benjamins,Amsterdam, 63-83
2002 Everaert, Martin The Case of the Theta System, Theoretical Linguistics 28, 325–
339.
2008 Evans, Nicholas. Reciprocal constructions: Towards a structural typology 33-104,
in : Ekkehard König and Volker Gast (eds), Reciprocals and Reflexives. Theoretical
and typlogical Explorations
2008 Everaert, Martin Domain restrictions on reciprocal interpretation, In: Ekkehard
König and Volker Gast (eds), Reciprocals and Reflexives. Theoretical and typlogical
Explorations, pp 557-576.
References
Amritavalli, R., ‘Lexical Anaphors and Pronouns in Kannada’, in B. Lust, K. Wali, J.
W. Gair and K. V. Subbarao (eds.), Lexical Anaphors and Pronouns in Selected South
Asian Languages: A Principled Typology (Berlin/New York: Mouton de Gruyter,
2000), 49–112.
Everaert, M. 1986. The syntax of reflexivization. Dordrecht: Foris.
Kazenin, K. I., ‘Verbal Reflexives and the Middle Voice’, in M. Haspelmath, E. König,
W. Oesterreicher and W. Raible (eds.), Language Typology and Language Universals
(Berlin: Mouton de Gruyter, 2001), 916–27.
Kevichusa, M. 2007. Apects of Angami (Tenydie) syntax. Doctoral dissertation,
University of Delhi.
Kevichusa, M. 1997. Relative Clause Formation in Tenyidie (Angami). M. Phil.
dissertation. University of Delhi
Page 27
27
Lalitha Murthy, B. & K.V. Subbarao, 2000. Pronouns and Lexical Anaphors in Mizo.
In: Lust et al (2000). 777-840.
Lust, B. J. Gair, K. Wali and K.V. Subbarao (eds.). 2000. Lexical anaphors and
pronouns in selected South Asian languages. Berlin: Mouton de Gruyter.
Minegeshi, M. & G. Murmu. 2001. Santali basic lexicon with grammatical notes.
ILCAA, Tokyo University of Foreign Studies, Tokyo.
Nagaraja, K. S., Korku Language (ILCAA, Tokyo: Tokyo University of Foreign
Studies, 1999).
Pangersenla. 2005. Issues in the Interpretation of Anaphors in Ao (Mongsen). M.Phil.
dissertation. University of Delhi.
Patnaik, M. & K.V.Subbarao.2000. An initial note on lexical anaphors and pronouns in
Juang. In: Lust, B. et al. (eds). 841-859.
Pelleitier, R. 1994. The syntax of predication: Case in Telugu reciprocal clauses. [Talk
delivered at the South Asian Languages Analysis (SALA) conference at the University
of Pennsylvania.]
Reinhart, T & T. Siloni. 2005. The lexicon-syntax parameter: Reflexivization and other
arity operations. Linguistic Inquiry 36.3. 389-436.
Sarju Devi, T & K.V. Subbarao. 2003. Reduplication and case copying: The case of
lexical anaphors in Manipuri and Telugu In: Linguistics of the Tibeto-Burman Area
25.2. Fall 2002, 47-72. Also in: Perspectives in Honor of P.J. Mistry. (eds.) R. Laury,
G. Mc Menamin, S. Okamoto, V. Samiian & K.V. Subbarao. New Delhi: Indian
Institute of Language Studies.
Subbarao, K.V. and B. Lalitha Murthy. 1997. Aspects of Mizo Grammar. Ms.
Subbarao, K.V., & B. Lalitha Murthy, 2000. Lexical Anaphors and Pronouns in
Telugu. In: Lust et al (2000).
Subbarao, K.V. (in press). South Asian Languages: A Syntactic Typology, Cambridge
University Press, Cambridge.
Page 28
28
Subbarao, K. V., U. R. Hakacham and M. Gogoi-Dutta Ms. 2004. A Typological
Grammar of Rabha.
Wali, K. 2000. Lexical anaphors and pronouns in Marathi. In: Lust et al (2000). 513-
574.