LANGUAGE VARIATION IN SPACE AND IN TIME

A social-dialectological approach to variation in the Transitive-Perfective Clause in Dialects

of Marathi

Sonal Kulkarni-Joshi
Deccan College, Pune.

Abstract
This paper addresses the theme of the workshop by providing a social-dialectological slant on
variation in language. I will begin with a brief overview of the central theoretical and
methodological tenets of the variationist approach to language. Two methodological off-
shoots of the variationist approach - socio-historical linguistics and modern dialectology - are
briefly introduced for examining synchronic variation in the NIA language, Marathi and its
implications for examining language change. The paper provides a description of variation in
case marking and agreement in the transitive-perfective clause in regional varieties of
Marathi, including Konkani and Ahirani. The data are drawn from an on-going
dialectological survey of Marathi at the Deccan College. The data are compared with
historical sources including Grierson (1905). It is often not possible to directly analyse
language change in space, but synchronic evidence in the form of areal variation substitutes
for the diachronic dimension. We will analyse the regional variation within the socio-
historical framework and argue that the variation is the result of both language-internal and
language-external factors.

1. Introduction
Social dialectology differs from traditional dialectology in shifting the focus from invariant,
archaic, rural forms of language used by settled communities to incorporating variationist /
sociolinguistic methods of sampling as well as the quantitative methods of analysis based on
data from large corpora (e.g. Siewierska and Bakker 2006).

Dialectology, a precursor of sociolinguistics, examines divergence of two local

dialects from a common ancestor and synchronic variation in the regional varieties.
Sociolinguists, on the other hand, are interested in the full range of forms in a community
(and their social evaluation). Sociolinguists use information about social structure, people
movements, extra-linguistic situation, contextual factors and social evaluation of structural
options in explaining mechanisms of language change / evolution. Modern dialectology
integrates a discussion of these social factors as also historical facts in the interpretation of
dialectal variation and change. Modern dialectology not only identifies the areal distribution
of particular linguistic features but also takes interest in the effect of mobility and contact
with speakers on the speech variety / varieties of a region.

Social Dialectologists believe that languages are inherently variable. Such variation is
not “free” but is “structured heterogeneity” (Weinreich et al 1968:188). Further, language
evolution is variational (like biological evolution), proceeding by competition and selection
among competing linguistic alternatives: A and B (and C), with A or B (or C, or A and C, or
B and C) prevailing because they were favoured by particular ecological factors (Mufwene
2001).
 The research agenda for studies of dialect / language variation and change was
charted by Weinreich, Labov and Herzog (1968) in their seminal paper, „Empirical
Foundations for a Theory of Language Change‟. This agenda can be summarised in the form
of five aspects of language change:

The constraints problem: The constraints problem involves formulating „constraints on the
transition from one state of a language to an immediately succeeding state‟ (Weinreich,
Labov and Herzog 1968:100).

The transition problem: This is the question of what intervening stages can (or must)
be posited between any two forms of a language separated by time. (Weinreich, Labov and
Herzog 1968:184).

The actuation problem: why the change was not actuated sooner, or why it was not
simultaneously activated wherever identical functional conditions prevailed. This is
paraphrased by Walkden in the Handbook of Historical Syntax as follows: “What factors can
account for the actuation of changes? Why do changes in a structural feature take place in a
particular language at a particular time, but not in other languages with the same feature, or in
the same language at other times?”

The embedding problem: “How are the observed changes embedded in the matrix of
linguistic and extralinguistic concomitants of the forms in question? (That is, what other
changes are associated with the given changes in a manner that cannot be attributed to
chance?)” (Weinreich, Labov and Herzog 1968:185).

The evaluation problem: How do members of the speech community evaluate the change in
progress?

Of the five, Weinreich et al recognised the actuation problem, “why did a particular change
occur at a particular place at a particular time” to be at the heart of a theory of language
change. Theories of language change differ in that they deal either with language-internal
factors (e.g. language acquisition, cognition, language use) or with language-external factors,
which concern population dynamics (e.g. migration / population movements, contact,
network ties, imperfect learning). The latter are examined by sociolinguists / social
dialectologists. The sociolinguistic approach to language variation and change (which
developed largely from the pioneering work of William Labov) includes consideration of
both linguistic constraints (e.g. the conditioning environment) as well as sociological and
contextual constraints (e.g. speaker‟s age, sex, education, formality etc.).

Social dialectology introduced sociolinguistic sampling methods to dialectology; data

are collected from a wide spread of speakers in the local speech community, including
speakers who are mobile and have come in contact with other regional speech varieties.
Speakers belonging to diverse age-groups, educational and professional backgrounds and
both sexes are sampled. (For an overview of applications of this method see Trudgill et al
2003.) The particular methodology helps to examine the mechanisms of diffusion of language
/ dialect change which can then be modelled (e.g. the cascade model or the gravity model,
Trudgill et al 2003).

Besides addressing traditional areas of sociolinguistic variation and change, social

dialectology is also concerned with newer areas of research such as dialect formation, dialect
diffusion and dialect levelling. These are the mechanisms by which language change is
effected.

Dialectology has forged interfaces with sub-disciplines other than sociolinguistics too.
In recent times there has been a growing realisation of the need for collaboration among
syntacticians and typologists on the one hand (who deal with cross-linguistic data drawn from
standard varieties; e.g. data presented in the World Atlas of Language Structures see
www.wals.info) and dialectologists / sociolinguists (who deal with non-standard, spoken
varieties; e.g. Linguistic Survey of India https://dsal.uchicago.edu/books/lsi/https:/ and
Romani Project /romani.humanities.manchester.ac.uk/). Sub-disciplines such as Syntax and
Typology are now turning attention to variation in language. Dialectology is seen as
complementing the typological interest in cross-linguistic variation by making available a
larger number of attested grammatical systems. A further advantage is seen in the dialects as
non-standardised grammatical systems (unlike the languages that typology generally deals
with). The advantage is that dialectal data gives typologists and syntacticians a larger number
of attested grammatical systems to explain within their theoretical frameworks. Dialectology
(whether regional or social) has focussed attention on non-standard speech varieties;
typological linguistics and syntax, on the other hand, have tended to focus attention on
standard languages. We are witnessing today a cross-fertilisation of methods from sub-
disciplines of linguistics - dialectology, historical linguistics, typology and contact linguistics
- in mutually beneficial ways (e.g. Bisang 2004; Chamoreau et al 2012 ). This development
has led to fresh opportunities for explaining language change using dialectological data.

However, the role of dialectology is often that of a hand-maiden (one which provides
rich dialectal data) just as it was in the nineteenth century for historical linguistics. A truly
fruitful integrated approach to language variation and change must accommodate the goals of
dialectology. Having identified the areal spread of a given structural feature, social
dialectologists seek answers to questions such as the following:

i. How did a particular regional variety come to have the linguistic features that it has?
ii. Do the optional structures x and y co-exist in an idiolect / dialect or is only one of the
structures possible in an idiolect? (i.e. is the variation inter-speaker or intra-speaker?)
iii. Are there systematic linguistic and social contexts in which either option / variant is
preferred by the speaker?

This paper will focus on (i) describing synchronic dispersion in the morpho-syntactic
feature of ergativity in the spatial domain in the Marathi-speaking region; (ii) comparing the
synchronic data with historical sources to draw indirect inferences about dialect change; (iii)
pointing to questions and generating hypotheses for further study of variation in space and in
time in the Marathi region. I will attempt to account for patterns of variation in the
geographical and temporal dispersion of ergativity within a usage-based framework which
draws on the sociolinguistic theory.

The remaining sections of the paper are organised as follows. Section two introduces
socio-historical linguistics as a methodology for examining variability in the spatial, temporal
and social domains. Section three is focussed on variability in the linguistic feature,
ergativity. Fresh dialectal data from regional varieties of Marathi is presented and compared
with specimens from the Linguistic Survey of India (1905). Optionality in regional as well as
in idiolectal usage will be described in order to raise relevant questions and generate
hypotheses for further examination within the framework of social dialectology.
2. Socio-historical linguistics: a methodological off-shoot
of variationism
The analysis of variable dialectal data in this paper employs two methodological off-shoots of
variationism : social dialectology and socio-historical linguistics. We will briefly describe
and illustrate these approaches before proceeding to addressing the main goals of the paper.
Socio-historical linguistics uses the quantitative, variationist methods of
sociolinguistics to examine diachronic development of social / regional dialects. A central
assumption of the approach being used is that the linguistic forces which operate today are
not unlike those of the past (Romaine 1982) i.e. there is no reason for assuming that language
did not vary in the same patterned way in the past as it does today (cf. the uniformitarian
principle). Current variation and its correlation with social structure and patterns of human
interaction may be used in constructing a social model. The approach helps the researcher to
investigate whether and to what extent synchronic variation in contemporary regional
varieties of a language reflects diachronic developments. (See Romaine 1982 for a case study
of syntactic variation in Scots English using the sociohistorical approach.) Methods such as
age-grading or apparent time are employed in making use of synchronic data to reconstruct
language change within a speech community (see e.g. Sankoff 2006.)

To illustrate the socio-historical methodology used to study language variation and

change, I reproduce below a case study of the transitive-perfective clause in the variety of
Marathi spoken in the border town of Kupwar (reported originally in Kulkarni-Joshi 2016).

Gumperz and Wilson (1971) was an influential study in the field of contact
sociolinguistics. They made a case for isomorphism or the development of identical syntactic
structures in the contact varieties of Marathi, Kannada and Hindi-Urdu in the town of
Kupwar located in the state of Maharashtra (where Marathi is the state official language)
close to the border with the state of Karnataka (where Kannada is the state official language).
Gumperz and Wilson presented data to suggest that close contact among the three speech
varieties over several hundred years had led to the putative syntactic isomorphism. Kulkarni-
Joshi (2016) used synchronic and diachronic data from Kupwar and the surrounding Marathi-
Kannada bilingual region at the state border to demonstrate that isomorphism was an artefact
of the particular methodology used by the researchers in the previous study. The socio-
historical approach and the apparent age construct were instrumental in arriving at this
conclusion.

A linguistic feature in the Kupwar variety of Marathi (A New Indo-Aryan language)

which was reported as affected by contact with Kannada (a Dravidian language) was the
syntax of the transitive-perfective construction. Gumperz and Wilson reported the loss of
ergativity in this speech variety under the influence of the non-ergative Kannada. Data
collected in the re-visit of Kupwar revealed that (i) ergative marking may be present or absent
on the subject NP of a perfective clause and (ii) the verb in such a clause may agree with the
subject NP which may or may not be case-marked or with the non-case marked object NP.
The analysis of agreement in the transitive perfective clause was based on the following
number of tokens (= instances of use of the transitive-perfective construction): Kupwar 58
tokens from 8 speakers; Hittani 62 tokens from 9 speakers; Bijapur 13 tokens from one
sample; Dharwar sample 1–11 tokens; Dharwar sample 2–5 tokens. Of these, the Kupwar
(District Sangli, Maharashtra) and Hittani (District. Belgaum, Karnataka) data were collected
by the author. Bijapur and Dharwar data are from Grierson (1905).

A comparison of the particular linguistic feature in Kupwar Marathi with other

contact varieties of Marathi close to the Maharashtra-Karnataka border (Fig. 1) showed that
the non-contiguous contact variety of Marathi in Dharwar showed a complete shift towards a
Kannada-like agreement pattern in transitive perfective clauses; some variability was
observed in the Bijapur sample; Kannada-like agreement was predominant in the Bijapur
sample. Data collected in the border village of Hittani (in the state of Karnataka) showed a
cross-generational pattern of variation identical to that in KuM. In both KuM and in Hittani
Marathi, standard Marathi-like agreement dominated. Quantification of the linguistic feature
relvealed that Kupwar Marathi and Kupwar Kannada had not reached complete
intertranslatability as was claimed by Gumperz and Wilson. Marathi-like agreement was
predominant in Kupwar Marathi across age groups. Younger speakers of KuM appeared to
have moved to a more standard Marathi-like usage. Cross-generational usage of the particular
linguistic feature in KuM was identical to that in the cross-border variety of Marathi in
Hittani village; Kupwar Marathi usage was, however, unlike that in contiguous Bijapur and in
non-contiguous Dharwar. Indirect inferences regarding dialect change could thus be drawn
using „age of speaker‟ as a social variable. This development was ascribed to the changed
status of Marathi in the region (i.e. that of state official language) since the formation of
linguistic states in 1960.

120

100

80

60
Variation
40
Marathi-like
20

Figure 1: Frequency of Marathi-like agreement in the transitive perfective clause in varieties

of Marathi spoken in Kupwar (south Maharashtra), Hittani (north Karnataka), Bijapur and
Dharwar (Karnataka).

The socio-historical method investigates both linguistic and social constraints on

language change (cf. the constraints problem identified by Weinreich et al 1968, see above).
In a study of dative marking in historical and dialectal varieties of Marathi, for instance, we
found that the change from the regional / traditional marker of the dative to the modern /
standard marker was constrained by semantic functions of the NP to which the marker was
attached: the modern marker first emerged to mark the recipient, beneficiary and goal
functions while the functions of purpose, external possessor, etc. were affected later in the
course of this development (see Kulkarni-Joshi and Kelkar forthcoming).

2.1 Limitations in using the socio-historical linguistic

approach
Reconstruction of diachronic changes in language using synchronic data within the socio-
historical approach relies significantly on having access to comparable data from different
points in time. Typically these are written texts. Very often the varying genres of the
available texts make it difficult to compare and draw reliable inferences about language
change. At times it is the formal nature of written language which renders texts a less than
satisfactory source of information. Marathi is fortunate to have literary representations of the
various phases in its historical development; however, a problem I have faced in my use of
these texts for applying the socio-historical method is that dialectal features which are typical
of spoken language are often not captured in the written texts. For example, the regional
variants of the transitive-perfective construction in Marathi are rarely found in writing. My
examination of this linguistic variable in dialects of Marathi (section 3 below) is therefore
largely restricted to comparison with specimens of Marathi in Grierson‟s Linguistic Survey of
India which is a written record of varieties of the language as it was spoken a little over a
hundred years ago.

We will now move to a description of ergativity in selected varieties of Marathi.

3.0 Variability in the transitive-perfective construction in

Marathi
Ergativity in Indo-Aryan languages is a well-studied phenomenon and has been examined
using various approaches in linguistics – historical, typological, syntactic, and sociolinguistic.
Previous studies of the ergative construction in western NIA include Bhatt, 2007; Davison,
2004; Kachru, 1987; Kachru & Pandharipande, 1978; Mahajan, 1990, 1997, 2012; Mohanan
1994; Subbarao, 2012, and others for Hindi-Urdu; Deo & Sharma, 2002; Pandharipande,
1997 for Marathi; Patel, 2007 for Kutchi Gujarati; Khokhlova, 2000, 2002 for Marwari;
Bickel & Yadava, 2000 for Nepali; and Bhatia, 1993; Bhatt, 2007; Butt and Deo, 2001;
Chandra, Kaur and Udaar, 2014 for Punjabi.

The focus of this paper will be on a closer examination of spatial and temporal
variation in the morpho-syntax of the transitive-perfective clause in Marathi. Present-day
standard Marathi is a split ergative language (split for person and aspect). Only third person
nominal expressions are marked for the ergative case and the verb in such clauses agrees with
the nominative object NP. Overt ergative marking is absent in the first and second persons
though the rule for agreement remains the same. Old Marathi (which is accessible in literary
and inscriptional texts) reveals that Old Marathi (c.1000 AD to 1390 AD) overtly case
marked the agent in all three persons and the verb agreement was with the nominative object
(or the verb took default neuter agreement). The regional varieties of present-day Marathi
reveal Old Marathi-like or standard-Marathi-like nominal marking and agreement pattern or
they have lost ergativity altogether; yet others show variability in marking the agent with
ergative case and variability in patterns of agreement as well. Such variability is more evident
in the speech varieties at the borders of the Marathi-speaking region today.
The synchronic, contemporary data for this study are drawn from five regional varieties of
Marathi:
i. the standard dialect based on the educated, Pune variety;
ii. the Nagpur variety bordering Hindi to the north-east of the Marathi-
speaking region today;
iii. Ahirani bordering Gujarati to the north-west of the Marathi-speaking
region today;
iv. the Sangameshwar speech variety in the Konkan close to Goan Konkani to
the south-west of the Marathi-speaking region today; and
v. southern variety of Marathi in Kolhapur district bordering Dravidian
Kannada.
The data were gathered in the course of an on-going Survey of the Dialects of the Marathi
language at the Deccan College, Pune 1. The focus of this project is on capturing morpho-
syntactic variability. Data were gleaned through personal narrations, narrations of traditional
stories and responses to a semi-structured questionnaire based on videos developed by the
project team at the Deccan College. Elicitation and translation could not have been useful in
collecting data on dialects of a single language. Further, the relatively infrequent occurrence
of morpho-syntactic variables in natural speech is well-known. Hence, videos were
developed to elicit particular agreement patterns, case markers, verb forms, etc. These
responses were cross-checked with data from narrations. Narrations are seen as advantageous
for employing the social-dialectological approach. Narration is a cultural universal, they have
ready accessibility, length of discourse specimens facilitates statistical counts (cf. quantitative
analyses in social dialectology) and it guarantees availability of a number of examples of
given construction-types in the text (Hopper Thompson 1980: 282). For the diachronic
dialectal data in this paper I have relied on published sources, mainly the Linguistic Survey of
India (1905; Vol. VII) and partly on Ghatage‟s Survey of Marathi Dialects (for Kudali).

3.1 Old Marathi

The transitive-perfective construction in Old Marathi had overt morphological marking on
personal pronouns in all three persons. The first and second person plural pronouns in Old
Marathi show syncretism. Agreement was with the nominative object. The available written
records for old Marathi suggest a homogenous, non-variable ergative system in Old Marathi.

1
The project is funded by the Rajya Marathi Vikas Sanstha of the Government of Maharashtra and is being
implemented at the Deccan College, Pune since September 2017. Till date, data have been collected in sixteen
of the 36 districts in Maharashtra from 154 villages and 1543 speakers.
ASPECT Person Number

Singular Plural

Non-Perf 1 mi amʰi

Perf 1 miya am ʰi

Non-Perf 2 tu tumʰi

Perf 2 tuwa tumʰi

Non-Perf 3 to tyani

Perf 3 tene tyani

Table 1: Pronominal paradigm of Old Marathi (based on Tulpule 1960)

Examples of the transitive-perfective construction in Old Marathi

1. mīya ramatẽ mhǝṇitlẽ [Līḷāčǝritrǝ: Līḷā 317]

I.ERG ram.ACC say.PFV.3SN

„I said to Ram.‟

2. tẽhĩ gosawīyãtẽ dekhilẽ [Līḷāčǝritrǝ: Līḷā 315]

he.ERG sage.ACC see.PFV.3SN

„He saw the sage.‟

3.2 The standard dialect of Marathi

Split-ergativity, where the ergative case marking occurs only in the perfective aspect, is
reported for the western NIA (New Indo-Aryan) languages such as Hindi-Urdu, Gujarati,
Punjabi, Sindhi, but not for Bangla, Oriya, Bhojpuri, and Marathi. In the standard dialect of
Marathi, only the third person subject NP of a finite transitive clause in the perfective aspect
bears ergative marker (i.e. the case marker -ne/-ni/-nə); direct objects and subjects of
intransitive clauses are nominative and the verb agrees with the non-case marked NP (3a); if
both subject and object NPs are case-marked, the verb shows default, neuter agreement (3b).
Overt morphological-marking on the first and second person pronouns in the ergative
construction of Old Marathi is lost in present-day standard Marathi. Further, the 1st and 2nd
person pronominal forms are identical to the corresponding non-perfective pronominal forms
in OM. Syncretism in the first and second person plural pronouns of OM persists in today‟s
standard variety.

Standard Marathi
3a. ti-nə/e/i kagəd phaɖ-l-a

she-ERG paper.3SM tear-PERF-3SM

She tore the paper.

3b. tyanə/e/i muli-la mar-l-e

he-ERG girl.3SF-ACC hit-PERF-3SN
He hit the girl.

When Marathi was codified / standardised towards the end of the nineteenth century, the
standard variety was based on the variety spoken by the educated elite class in Pune (Poona).
Grierson‟s data for Poona Marathi reveals variability in the forms of the first and second
personal pronouns in a single speaker‟s speech (myã and mi in competition; twã of Old
Marathi is still retained in Poona variety of the late nineteenth / early twentieth century.) This
variability is absent in present day standard / Pune Marathi.

Poona Marathi of the late nineteenth century (Grierson 1905: Poona Specimen 1)
4. tujhi adnya mi kadhĩ-hi moḍǝli nahĩ
your.SF command.SF I.ERG when-even break.3SF NEG
„I never disobeyed your command.‟

5. tǝri myã apǝlya mitra-bǝrobǝr čǝin kǝrawi mhǝṇūn twã mǝla kǝdhi kǝrḍũ hi dilẽ nahĩs
yet I.ERG self.OBL friend.OBL-with revelry.3SF do.SUBJ therefore you.ERG I.DAT
when young lamb even give.3SN NEG.2S
„Yet (you) never gave me a lamb‟s young one so that I could make merry with my friends.‟

We will examine verbal agreement and ergative marking on the subject NP in the transitive,
perfective clause in regional varieties of Marathi. Our interest will be in questions such as the
following: What is the nature of variability observed in the morpho-syntax of the transitive-
ergative clause in contemporary regional varieties of Marathi? Can we trace the trajectory of
change from Old Marathi-like system / a historically prior state to the present-day? What are
the socio-linguistic / socio-historical correlates of the observed change(s)?

3.3 The dialectal data: diachronic and synchronic

The SDML (Survey of Dialects of the Marathi Language) data collected so far indicates three
broad patterns in the morpho-syntax of the transitive-perfective construction (listed below as
A, B and C).

[A] Retention of Old Marathi-like ergativity in Nagpur Marathi

Of the surveyed regional varieties, the Nagpur variety retains overt morphological marking
on personal pronouns in all three persons and agreement with the nominative object in the
transitive-perfective clause, as in Old Marathi.
ASPECT Person Number

Singular Plural

Non-Perf 1 mi ami

Perf 1 mya, mi ami

Non-Perf 2 tu tumi

Perf 2 twa tumi

Non-Perf 3 to/ti/te ??

Perf 3 tyanə tyani

Table 2: Pronominal paradigm of the Nagpur dialect (based on LSI data)

ASPECT Person Number

Singular Plural

Non-Perf 1 mi ami

Perf 1 mi/miya/minə ami

Non-Perf 2 tu tumi

Perf 2 tu/tya/tunə tumi

Non-Perf 3 to/ti/te NA

Perf 3 tyani / tini / tyani

tyanə
Table 3: Table Pronominal paradigm of the Nagpur dialect (based on SDML data)

Example of Nagpur Marathi of the late nineteenth century (Grierson 1905: Specimen no. 1)
6. mya apǝlya mitra-bǝrobǝr čǝin kǝray-saṭhi twa mǝ-la kokǝru dekhil dellǝ nahi
1S.ERG self.OBL. friend-with fun do.NON.FIN.-for you.ERG 1S-DAT young goat.SN
even give.PFV.3SN NEG
„You didn‟t even give me a young goat for me and my friends to play with.‟
Example of Nagpur Marathi of the present times (SDML, 2019)
7. mi-nǝ khal-un pay-l-ǝ
I-ERG below-ABL see-PFV-3SN
„I saw (something) from below.‟
The past tense is formed as in the Dekhan,; thus mya marǝlǝ I struck, tya marǝlǝ thou
struckest. The third person singular of transitive verbs sometimes ends in ǝn; thus dhaḍǝlǝn
sent. […] The past tense of transitive verbs is used in the same way as in the Dekhan, the
subject being put in the case of the agent, and the verb agreeing with the object in gender and
number or being put in the neuter singular (LSI Vol. VII, p. 221).
We note an expansion in terms of morphological forms available in the first and
second person singular pronouns in the transitive-perfective clause in the Nagpur speech
variety. The predominant forms heard in the course of field work in rural Nagpur were minə
„I-ERG” and tunə ‘you-ERG‟. Interestingly, these are not attested in the LSI Nagpur
specimens. (Hindi-like ergative pronominal forms marked by –ne are attested in the LSI
further to the east in the Chhindwara speech variety (see Grierson 1905: 319-329). Further
research would reveal the developments which led to the present-day forms in the Nagpur
dialect. [Nagpur and Chhindwara were a part of the Central province under British rule.] Was
a non-local morphological marker adopted to mark the ergative in the Nagpur variety? If yes,
how do we account for the observation that a non-local feature was adopted by this regional
variety of Marathi? Or, could the –ni marking in the first and second persons have resulted
from an analogical change modelled on the third person pronouns? Answering these
questions requires further examination of the data.

Agreement pattern in the Nagpur speech variety, both in the LSI and in the SDML
data are as in standard Marathi: the verb agrees with the non-case marked object NP or shows
singular neuter agreement.

[B] Development of standard-Marathi-like ergativity: Ahirani

Ahirani (also referred to in the literature as Khandeshi, Dhed Gujarai) instantiates this type of
ergativity among the dialects of Marathi. Ahirani (Khandeshi) is not classified as a dialect of
Marathi by Grierson, but as belonging to the central NIA group along with Bhili, Banjārī or
Labhānī, Bahrūpiā, etc. Grierson provides two specimens of the speech variety sampled in
Khandesh, Nandurbar taluka (Grierson 1905, Vol IX.3. specimen nos. 65 and 66). The
pronominal paradigms for Ahirani of the late nineteenth century (LSI data) and that of the
present day are presented in Tables 4A and 4B respectively.

ASPECT Person Number

Singular Plural

Non-Perf 1 mi, mǝi ham, am, apǝn

Perf 1 mi, me ami, amhu

Non-Perf 2 tu tum

Perf 2 tu, tuna tumi, tumhi

Non-Perf 3 to, ti, te te, tya

Perf 3 tyane NA

Table 4A: Pronominal paradigm of the Khandeshi/ Ahirani dialect (based on LSI Vol. IX.3
data, p. 209)

ASPECT Person Number

Singular Plural

Non-Perf 1 mi ami, amhi

Perf 1 mi ami, amin, amhi

Non-Perf 2 tu tumi

Perf 2 tu tumi, tumin

Non-Perf 3 to/ti tya

Perf 3 tyani tyasni

Table 4B: Pronominal paradigm of the Ahirani dialect (based on SDML data)

A caveat we would like to add immediately is that the comparison of the ergative
constructions in Ahirani in the LSI and in the SDML is being attempted here although the
LSI specimen was collected in Nandurbar district and the available SDML specimen was
collected in the neighbouring Dhule district.
On comparing the two paradigms, we note complete syncretism in the first and second
person singular forms in present-day Ahirani; the forms mi and tu are used both in non-
perfective and in perfective constructions. This development in Dhule Ahirani may have
resulted from close contact with / bilingualism in Marathi.
 We do not see similar syncretism in the first and second person plural pronouns.
Ahirani is said to represent a grammatical system having a mix of characteristics of
neighbouring Gujarati and Marathi. But we find that the plural forms in the pronominal
paradigm of Ahirani are more differentiated than those of either Marathi or Gujarati (cf.
section 3.2 for standard Marathi and Table 5 for Gujarati). While the nominative and agentive
first and second person plural pronouns in both Marathi and Gujarati show syncretism, the
equivalent pronominal forms in Ahirani (LSI Vol. IX.3, p. 209 as well as SDML data) show
absence of syncretism. However, the SDML Ahirani data shows overlap among the
pronominal forms used in non-perfective and in perfective constructions. This variability
reflects reflexes of diachronic change and indicates a period of fluctuation and potential
language change. It will be interesting to note the projected direction of this on-going change.
The –n marking on first and second person plural pronouns in the perfective (amin we.ERG,
tumin you.PL.ERG) may also have resulted from analogy with the third person pronominal
forms. Educated Ahirani speakers optionally used standard-Marathi-like pronominal forms.

Verbal agreement in the LSI specimen of Khandeshi / Ahirani, and that in the present-
day Dhule Ahirani (SDML) is like that in Old Marathi and in Standard Marathi.
8. tya-ni tyas-le apǝli jinǝgi waṭ-ī did-ī
he-ERG he-DAT self.3SF property.SF distribute-CP give.PFV.3SF
„He divided his property (among his sons).‟
(LSI Vol IX.3, Specimen No. 65 collected in Nandurbar district)

9. te por-ni tya manus-le piwan paɳi di-n-ə

that girl-ERG that.OBL man-DAT drink.NON.FIN water.N give-PFV-3SN
„The girl gave the man water to drink.‟
(SDML data collected in Dhule district)

10. amin khir khadi

we.ERG*2 porridge.SF eat.PFV.3SF
„We ate the porridge.‟
(SDML data collected in Dhule district)

ASPECT Person Number

Singular Plural

Non-Perf 1 hu ame, am

Perf 1 mẽ ame

Non-Perf 2 tu tǝme, tam

2 st
The nominative and ergative forms of 1 plural need to be cross-checked.
Perf 2 tẽ tǝme

Non-Perf 3 te teo

Perf 3 teṇe teoe

Table 5: Pronominal paradigm of Gujarati (based on LSI data, Vol. IX.2)

[C] Reduction of ergativity: Konkani in Ratnagiri

In this sub-section we will consider the Sangameshwar speech variety spoken in the Ratnagiri
district in south-west Maharashtra. 3

ASPECT Person Number

Singular Plural

Non-Perf 1 mi əmi

Perf 1 mini əmi

Non-Perf 2 tũ tumi

Perf 2 tuni tumi

Non-Perf 3 to / ti te

Perf 3 tyani/tini tyanni

Table 6. Pronominal paradigm of the Sangameshwar speech variety (based on LSI data; p.66)

ASPECT Person Number

Singular Plural

Non-Perf 1 mi əmi

Perf 1 mi əmi

3
Grierson makes a difference between the Konkan standard (which includes varieties such Agri, Bankoti and
Sangameshwari in the coastal stretch from Thane to north Ratnagiri) and Konkani spoken in the region
extending from Rajapur in Ratnagiri district uptil Sindhudurg district.
Non-Perf 2 tu tumi

Perf 2 tu tumi

Non-Perf 3 to / ti te

Perf 3 tyane/tine tyanni

Table 7. Pronominal paradigm of the Sangameshwar speech variety (based on SDML data)
We compared the pronominal paradigms of the Sangameshwar speech variety as attested in
the LSI and in the SDML; over time, syncretism has developed in the first and second person
singular pronouns. The LSI evidence shows that, unlike Old Marathi, case syncretism
occurred in the first and second person plural pronouns in Sangameshwar a hundred years
ago.

Grierson (1905: 122) observes that Sangamshwari closely agrees with the Konkan
Standard of Marathi (cf. footnote 4). Verbal agreement in the transitive–perfective clause in
Sangameshwar variety / Standard Konkan is with the object, even if it is inflected (as in Goan
Konkani and in Gujarati). See examples 11-15 below: sentences 12-15 show agreement with
the case-marked object; sentence 15 shows agreement with a non-inflected object NP vãṭṇi.

11. tya giresta-n hya-s ḍukrã tsaraya-s seta-var dhaḍ-l-an

that person-ERG he-DAT pigs feed-DAT field.OBL-LOC send-PFV-3S
„That person sent him to the field to feed pigs.‟
(LSI 1905:125)

12. bapǝsa-n tya-s [….] miṭi marǝ-l-an

father-ERG he-DAT […] embrace.SF hit-PFV-3S
„Father embraced him.‟
(LSI 1905:126)

13. tya-nǝ eka gəḍ-ya-s sad ghət-l-an

he.OBL-ERG one.OBL labourer-OBL-DAT call.SF throw-PFV-3S
„He called a labourer.‟

14. leka-n bapsa-s pərət bolna ke-l-an

child-ERG father-DAT again speech.SN do-PFV-3S
„The child once again spoke to the father.‟

15. mǝg tya-nǝ tyãs-ni apǝlya jinǝgi-č-i vãṭṇi

then he-ERG he-DAT self.OBL property.OBL-GEN-3SF division.SF

kǝrun di-l-i
do-CP give-PFV-3SF
„Then he divided his property and gave (his son) his share.‟
(LSI 1905:125)

In data collected for the SDML project, we find variability in verbal agreement in the
Sangameshwar villages. The variability appears to be correlated with the religion of the
speaker. Non-Muslim collaborators in Sangameshwar use standard-Marathi-like agreement
(with non-inflected object NP). Among Muslim collaborators in Sangameshwar (Karanjari,
Amavali, Kondivare and Kasba villages), the verb agrees with the second person plural and
third person (sg. or pl.) subject; elsewhere the verb agrees with the nominative object (Kazi
2019). It is also worth noting that non-Muslims in Sangameshwar / Ratnagiri report Marathi
as their home language while Muslims report Kokni to be the home language (ibid.)

16. tu -ya ḍæḍi-ni tu- ya pəppa-ni

you-GEN-OBL daddy-ERG you-GEN.OBL father.OBL-ERG

parṭi ṭhəw-len həyt

party.SF keep-PFV.3PL be.PRS.3PL
„That’s why your father has organized a party.‟
(Excerpt from the Prodigal Son Story collected for the SDML from Muslim collaborator in
Kasba village by Kazi 2019)

In sentence 15 (collected in the Kasba village in Sangameshwar taluka), the verb ṭhəw-len
agrees with the ergative subject dædi-ni or pəppa-ni (3rd HON.). Similarly, in the sentences
tumi kagəd phaḍlew „You (pl) tore the papers‟, tumi porala nijəwlew „You (pl) put the child
to sleep.‟ the verb agrees with the subject. 4

[D] Loss of ergativity in southern Maharashtra

In this sub-section we Chandgad-Gadahinglaj
In the southern variety (both the Gadahinglaj and Chandgad varieties) of Marathi spoken in
the district of Kolhapur, overt ergative marking is present in all three persons (as in Old
Marathi but unlike the contemporary standard variety). The actual pronominal forms differ:
Chandgad variety has the forms mya (1st p.)and tiya (2nd p.) while the Gadahinglaj variety has
minǝ (1st p.) and tunǝ (2nd p.). In both varieties these pronominal forms co-vary with the
forms mi (1st p.) and tu (2nd p.) of the standard speech variety. A further similarity between
the transitive-perfective clauses of the two southern varieties is preferred verbal agreement
with the subject NP which is unlike all other regional varieties of Marathi. The verb agrees
with the person and gender of the subject NP.

ASPECT Person Number

Singular Plural

Non-Perf 1 mi amʰi

Perf 1 mi/miya amʰi

4
Deo and Sharma (2009) make the following observation for the Gowari dialect of the documented in Grierson
(1905): “transitive perfective subjects in the first and second person do trigger agreement, suggesting that they
are behaving like nominatives in both morphology and abstract case features”.
Non-Perf 2 tu tumʰi

Perf 2 tu/tiya tumʰi

Non-Perf 3 to/ti/te tyani

Perf 3 to/tyanə tyani

3 ti/tinə tyani

Table 8: Pronominal paradigm of the Gadahinglaj speech variety (based on SDML data)

In sentence 17, the subject NPs are not marked with instrumental marker in perfective aspect
and the verb „to do‟ agrees with subject.
17. hi BA kelin, natu pǝndrawi kelyan
she BA do-PFV-3SF grandson 15th do-PFV-3SM
„She (grand-daughter) has completed graduation and the grandson has studied until
the 15th class.‟
In sentences 18 and 19, the subject is marked with instrumental marker in perfective aspect
and the verb agrees with subject.

18. tyani baṭli anlelyani

they.ERG bottle.SF bring.PFV.3PL
„They brought the bottle.‟

19. donʰi paḍlya baṭlya tyanə

both fall.CAUS.PFV.3PL bottle.3PL.F he-ERG
„He dropped both the bottles.‟

This variant of the transitive-perfective construction is observed in most varieties of Marathi

along the Marathi-Kannada border (see also Kulkarni-Joshi 2016 for data on Kupwar). The
subject NP forms appear to be marked with the case marker –ne / ni. But the subject
agreement prompts us to argue that the seemingly case-marked ergative subject NP has been
re-analysed as nominative in the Marathi-Kannada contact region. Arguably, the contact with
a Dravidian language, Kannada has triggered this reanalysis in the southern varieties of
Marathi.

20. tenǝ punyasnǝ alay (SDML data collected in Gadahinglaj)

he Pune-from come.PFV.PRST.3SM
„He has come from Pune‟

21. minǝ pepǝr wacayloy (SDML data collected in Gadahinglaj)

I newspaper read.PROG.PRST.1SM
„I am reading the newspaper.‟
 22. moṭhya hɔspiṭǝlat kam kǝruca ǝsǝ ṭhǝrǝvi-l-o mi (SDML data collected in Chandgad)
big.OBL. hospital.LOC work.SN do-PRED thus decide-PFV-1SM I
„I decided to work in the big hospital.‟
„

Our proposal that the ergative case-marked pronominal form has been reanalysed as
nominative is strengthened by the occurrence of a form such as teni , used as an honorific , a
pattern which seems to be modelled on the Kannada construction – tǝnde awǝru bandidu
(father he=HON. come.IMPF). The equivalent sentence in this variety of Marathi is presented
in (22).

23. pǝppa teni alyat (SDML data collected in Gadahinglaj)

father he (=HON.) come.PFV.PRST.3PL
„Father has come.‟
This section focused on describing broad patterns in the transitive-perfective clause in
selected regional varieties of Marathi.

3.3 Inter-speaker and intra-speaker Variability

Idiolectal (intra-speaker) variation too is attested in the SDML data. For instance, in the data
collected in southern Maharashtra in Gadahinglaj taluka of Kolhapur district (village Hebbal
Jaldyal) the following variability was observed in the speech of a female speaker aged 55
years:

24. mi-nə ajpəryet he diwəs bəɡitlə

I-ERG today-until these days see-PFV-3SN
„I saw these days until today.‟

25. ti enə kaḍʰun mi-nə pəi e bʰaɡiw-l-ə

those.F cow dung.PL remove-CP money I-ERG settle-PFV-3SN
„I picked cowdung to earn money and settle (these expenses).‟

26. hyo ḍo ɡər ewḍʰa ubʰa kela mi-nə

this.3SM mountain.SM. this much stand up do-PFV-3SM I-ERG
„I raised (such a big) this mountain.‟

27. mulɡya-la mi-nə kay mʰəṭlə

son.OBL-ACC I-ERG what said-PFV-3SN
„What I said to my son was that […].‟

28. mʰənun he kə ṭə kər-un mi-nə lokan-cə an ɡʰaṇ kadʰun mi-nə jətən ke-l-əy
hence these effort.PL do-CP I-ERG people-GEN cow dung dirt remove-CP I-ERG
preserve do-PFV-PRST
„I have done dirty jobs in order to save (money).‟

29. mi kaḍʰloy ki tu kaḍʰələs

I remove-PFV-1S-PRST or you remove-PFV-2S-PFV
 (He said to him), have I taken it out or have you taken it out.

30. mi tewḍʰə eṇ kʰa-ll-ə səɡḷ-ə

I that much.3SN cow dung..SN eat-PFV-3SN all.3SN
‚I took the blame.„
The occurrence of the first person singular pronominal forms mi and minǝ in the speech of
this single speaker was tabulated as in Table 9.

Perfective Clause Imperfective Clause

Transitive verb Intransitive verb Transitive verb Intransitive verb

minǝ 11 0 1 0

mi 14 6 9 10

Table 9: Idiolectal variation in Hebbal Jaldyal (Dist. Kolhapur) Total number of tokens
analysed = 51
It was observed that the presence or absence of the ergative marking on the pronoun is
contingent upon the clause type (perfective or imperfective) and verb type (transitive or
intransitive).
Samples collected from younger, educated speakers in this village show a complete
absence of the structural alternative „minǝ‟. Evidence for inter-speaker variation becomes
evident especially on comparing the speech of older speakers with that of younger, educated
speakers. We noted a strong preference for subject agreement among all speakers in this
region. Yet, among the younger speakers we see a shift towards standard Marathi-like object
agreement (31):

31. mi tu-la don kuraḍ-i dak-əw-l-ya (Female, educated speaker aged 25)
I.ERG you-DAT two axe.F-PL show-CAUS-PFV-3Pl.F.
„I showed you two axes.'
We summarise in the next section our main findings for the morpho-syntax of the transitive-
perfective construction in the selected regional dialects of Marathi.

4.0 Discussion and conclusion

The primary aim of this paper was to demonstrate the social dialectological approach to
variability in language. This version of dialectology was distinguished from traditional
dialectology. Socio-historical linguistics was introduced as a methodological off-shoot of
sociolinguistics; the method involved the use of quantification and correlation in comparing
synchronic stages of a language and drawing inferences about language change. This was
illustrated with the case of regional varieties of Marathi.

We examined the morpho-syntax of the transitive-perfective construction in selected

regional varieties of Marathi. The selected dialects included varieties spoken at the extreme
ends of the Marathi-speaking region today and the standard dialect. We noted variation across
these dialects both in case-marking the agent and in verbal agreement. The variation is
summarised in table 10.

Old Standard Nagpur Ahirani Sanga- South

Marathi Marathi meshwar Kolhapur

Nominal 1,2,3 sg 3 sg, pl 1,2,3 sg 1pl 3 sg, pl 1,2,3

Domain persons (Not persons 2pl (reanalysi
(Marking overt on 3 sg, pl s)
on Subj. 1,2
NP) persons)

Verbal object object object object object & subject

Domain subject
(agreeme
nt)
Table 10. Summary of ergative marking on the subject NP and verbal agreement in the
transitive-perfective clause across dialects of Marathi (The numbers 1, 2 and 3 indicate first,
second and third person pronouns.)

We noted variability at the community level, at the inter-speaker as well as intra-speaker

levels. Quantification of data collected in the course of the SDML project (using the
methodology demonstrated in section 3.1 for Kupwar Marathi) within the framework of
socio-historical linguistics will enable us to trace the diverse trajectories of dialect change. At
this stage in the SDML, the focus is more on identifying macro-patterns in the geographical
distribution of variants. Micro-level analysis, e.g. quantification of multiple tokens of the
same linguistic feature across speakers in a region, has not been attempted yet in the project.
However, certain tendencies are evident in the regional variation in the data. Both contact
with neighbouring languages and analogy with existing patterns in the dialect have played
roles in determining the structures of particular dialects. For now, we must leave the
identification of these facilitating / constraining factors to future research.

Since the key idea of this workshop was to find meeting ground for the functionalists
and formalists to examine variation in language, we conclude the paper by offering
suggestions for such collaboration. (Suggestions on how the research objectives of social
dialectology need to be incorporated in such an integrated research programme were listed in
the Introduction.) A social dialectological approach and formal approaches to language
variation differ in their assumptions about the nature of language, their goals and their
methods. While the former assumes that “Language is inherently variable” the latter relies on
the construct of the “ideal speaker-listener” and views variability largely as belonging to the
realm of „performance errors‟. Social dialectologists are involved in building probabilistic
models with some predictive value. They focus on identifying norms of language use shared
by the community. They recognise that languages / dialects change as a result of changed
sociolinguistic circumstances as well as factors such as evaluation of the dialect by its
speakers. They take up studies of both inter- and intra-systemic dialectal differences; these
could benefit from formal theories in linguistics in the following ways. Linguistic theory can
inform the decisions underlying the selection of dialect features to be studied, although the
selection will typically not solely be based on considerations of a strictly linguistic nature. As
suggested by Chambers and Trudgill (1998: 33), linguistic analysis can counteract the
“atomistic” approach to dialect features that is typical of dialectology in that its practitioners
have had a tendency “to treat linguistic forms in isolation rather than as parts of systems or
structures”.

References
Bisang, Walter. 2004. Dialectology and typology – An integrative perspective. B. Kortmann
(ed.), Dialectology Meets Typology. Dialect Grammar from a Cross-linguistic
Perspective (Berlin: Mouton de Gruyter), pp. 11–45.
Chambers, J. and P. Trudgill. 1998 (1980). Dialectology. Cambridge: CUP.
Chamoreau, Claudine and Isabelle Léglise (eds) 2012: Dynamics of Contact-Induced
Language Change (Language Contact and Bilingualism Series 2). Berlin and Boston: De
Gruyter.
Grierson, George. 1905. Linguistic survey of India. Vol. 7. Indo-Aran family. Southern
group. Specimens of the Marathi Language. Calcutta. Reprinted 1968. Delhi: Motilal
Banarsidass.
Grierson, George. 1907. Linguistic survey of India. Vol. 4.3. Indo-Aryan Family. Central
Group. The Bhīl Languages, including Khāndēśī, Banjārī or Labhānī, Bahrūpiā, &c.
Calcutta. Reprinted 1968. Delhi: Motilal Banarsidass.
Kazi, S. 2019. The Grammatical Sketch and Texts of Sangameshwari. Unpublished M.A.
dissertation submitted to the Deccan College (Deemed University), Pune. (Research
funded under the project Survey of Dialects of the Marathi Language (SDML).)
Kulkarni, S. 2001. Sociolinguistic variation among Marathi-speaking adolescents in Pune.
Unpublished PhD dissertation submitted to the University of Reading, UK.
Kulkarni-Joshi, S. 2016. „Forty Years of Language Contact and Change in Kupwar: A
Critical Reassessment of the Intertranslatability Model‟. 2016. International Journal of
South Asian Languages and Linguistics, De Gruyter, pp. 147-174.
Kulkarni-Joshi, S. 2017. „Substratum Effect and the Dravidian Element in Marathi: Towards
an Alternative Model‟. 2017. International Journal of Dravidian Linguistics Vol.
XLVI, No.2: 18-54.
Mufwene, Salikoko. 2001. The Ecology of Language Evolution. Cambridge: Cambridge
University Press.
Sankoff, G. 2006 Age: Apparent time and real time. Elsevier Encyclopedia of Language and
Linguistics, Second Edition, 2006. Article Number: LALI: 01479. Accessed online on
29 April 2019.
Siewierska, Anna and Dik Bakker 2006. Bi-directional vs. uni-directional asymmetries
in the encoding of semantic distinctions in free and bound person forms. In Types of
variation : diachronic, dialectal and typological interfaces. Nevalainen, T., Klemola, J.
& Laitinen, M. (eds.). Amsterdam: John Benjamins, p. 21-50.
Trudgill, P., D. Britian and J. Cheshire. 2003. Social Dialectology. John Bejamins.
Tulpule, S.G. 1949. Prachiin Marathi Gadya. Pune: Venus Book Stall.
Tulpule, S.G. 1963. Prachin Maraṭhi koriv Lekh. Pune: Pune University Press.
Tulpule, S.G. (ed.).1966. Līḷācharitra (purvārdha, Part 2). Nagpur-Pune: Suvichar Prakashan
Mandal.
Tulpule, S.G.1973. (2nd edition) Yadavkalin Maraṭhi Bhaṣa. Pune: Venus Prakashan.
Walkden, George. 2017. The actuation problem. In Cambridge Handbook of Historical
Syntax, eds. Adam Ledgeway and Ian Roberts. Cambridge: Cambridge University Press.
Weinreich, Uriel, William Labov, & Marvin I. Herzog. 1968. Empirical foundations for a
theory of language change. In Winfred P. Lehmann and Yakov Malkiel (eds.), Directions
for historical linguistics: a symposium, 95–195. Austin: University of Texas Press.

Pending References -

Bhatt, 2007;
Davison, 2004;
Kachru, 1987;
Kachru & Pandharipande, 1978;
Mahajan, 1990, 1997, 2012;
Mohanan 1994;
Subbarao, 2012, and others for Hindi-Urdu;
Deo & Sharma, 2002;
Pandharipande, 1997 for Marathi;
Patel, 2007 for Kutchi Gujarati;
Khokhlova, 2000, 2002 for Marwari;
Bickel & Yadava, 2000 for Nepali; and
Bhatia, 1993;
Bhatt, 2007;
Butt and Deo, 2001;
Chandra, Kaur and Udaar, 2014 for Punjabi.