** NOTE: This is a pre-final version. The published version of this paper can be found
in:
- Cardoso, Hugo C. & Patrícia Costa. 2021. ‘Synchronic variation in Sri Lanka
Portuguese personal pronouns’. Journal of Pidgin and Creole Languages 36(1): 81-113.
<https://www.jbe-platform.com/content/journals/10.1075/jpcl.00070.car>
--Synchronic variation in Sri Lanka Portuguese personal pronouns1
Hugo C. Cardoso and Patrícia Costa
Universidade de Lisboa – Faculdade de Letras, Portugal
Abstract
This paper presents and discusses the instances of synchronic variation attested in the
personal pronoun paradigm of modern Sri Lanka Portuguese, an endangered Portuguesebased creole spoken by relatively small communities scattered across Eastern and
Northern Sri Lanka. Although Sri Lanka Portuguese has a long history of documentation
dating from, at least, the beginning of the 19th century, only a few studies have explicitly
reported cases of synchronic variation. This study aims, therefore, to fill that gap, by
contributing to the description and explanation of patterns of variation relating to the
personal pronoun paradigm as encountered in documentary data collected between 2015
and 2020, over several field trips to the districts of Ampara, Batticaloa, Jaffna, and
Trincomalee. The nature of the variation observed in the data ranges from phonetic
alternations to strategies of paradigm regularization and stylistic shrinkage, often
revealing the effects of diachronic processes of variant competition and substitution.
Combining the observed patterns of variation with surveyed linguistic trends of language
shift, we propose that obsolescence may be responsible for some of the variability
encountered in modern SLP personal pronouns, especially that associated with certain
socially- or geographically-defined subsets of the speech community (viz. the younger
generations and the speakers from Jaffna) characterized by advanced language loss.
1
This study is the result of research activities supported by the Endangered Languages Documentation
Programme (grant: MDP0357), the Fundação para a Ciência e a Tecnologia (grant: IF/01009/2012), and
the University of Lisbon (through a doctoral scholarship attributed to Patrícia Costa in 2018). The authors
would like to thank the many Sri Lanka Portuguese speakers who contributed generously and patiently to
our documentation of the language, as well as Donald Winford and the editors of this special issue, Isabelle
Léglise, Bettina Migge, and Nicolas Quint. We are also thankful to the participants of the 2019 joint
ACBLPE and SPCL meeting in Lisbon and to three anonymous reviewers whose comments helped us
improve our study significantly. Finally, we wish to acknowledge Mahesh Radhakrishnan and Rui Pereira,
fellow members of the Documentation of Sri Lanka Portuguese project team, who collected and made
available to us some of the relevant data for this study.
1
Keywords: Sri Lanka Portuguese, personal pronouns, variation, endangered languages,
language obsolescence.
1. Introduction
While sociolinguistic considerations, by necessity, are never far from the study of creole
languages – be it because of questions related to their formation and development,
endangerment, coexistence with other languages, or others –, studies of communityinternal variation among these languages are not equally distributed. As Sippola
(2018:96) pointed out, sociolinguistic studies of variation have been less common for
Romance-based than for English-based creoles, and those that do exist have tended to
focus on some of the creoles with the largest numbers of speakers. The Portugueselexified creoles of South Asia, currently spoken by relatively small communities, are a
group of languages which have been somewhat neglected in this respect, even if
descriptive works occasionally report instances of synchronic variation and a few studies
have explicitly tackled such cases (e.g. Clements 1990; Cardoso 2007; see Cardoso 2014
for an overview and discussion).
This article contributes towards that descriptive effort by exploring the various
instances of variation concerning personal pronouns in modern Sri Lanka Portuguese
(henceforth, SLP), the Portuguese-lexified creole which formed in the 16th century on the
island then known as Ceylon and, though endangered, continues to be spoken in a few
discrete pockets in the Eastern and Northern parts of Sri Lanka. We base this study on a
recently-collected corpus of SLP (Cardoso 2017), part of an ongoing documentation
effort (see Cardoso et al. 2019), which records considerable variation in various domains
of modern SLP, well beyond pronouns.
This study of personal pronouns is a step towards systematizing the variation
encountered in these documentary materials and ensuring that the description of SLP
takes stock of it. Several authors have highlighted not just the mutual advantages but also
the challenges of combining language documentation and sociolinguistic analysis
(Meyerhoff 2017), and of making sure grammatical descriptions do not gloss over
variation (Nagy 2009). Potential issues arise from traditional differences in the scope and
methodologies of variationist research, on the one hand, and of language documentation
and language description, on the other. These challenges are amplified in the case of
under-documented or under-described languages – as opposed to languages with a
considerable history of research, for which large datasets may already be available and in
which patterns of variation may already have been identified –, since the identification
and interpretation of patterns of variation often has to proceed in tandem with the
2
delineation of the language’s basic grammatical properties (Meyerhoff 2017); and
especially so if the language is endangered, in which case there may be the additional
challenge of a small pool of speakers from whom to obtain linguistic data (Nagy 2017).
Mansfield & Stanford (2017:119) point out that the practical difficulties faced by outsider
researchers interested in making sense of how a language they are documenting interacts
with sociolinguistic variables derive from three factors: a) the fact that they are “cultural
outsiders” (which poses challenges to the very collection of reliable and representative
data); b) the fact that they must identify “variables with limited prior knowledge” (since,
when dealing with under-described languages, the incremental nature of knowledgebuilding is especially evident); and c) the fact that they must conduct “sociolinguistic
analysis with limited data”. We are aware that all of these apply in the case of our study
– although, as we explain in section 2, our analysis of SLP personal pronouns could rely
on prior descriptions –, which justifies the cautious nature of some of our conclusions,
but still allows progress in making sense of modern SLP variation.
Our first objective in this study, then, is to report on the full range of variants
within the SLP personal pronoun domain that can be identified in the corpus, in terms of
their form and function, much of which is absent from the previous literature. We will
furthermore attempt to identify any patterns in the distribution of variables which may
provide insights into the language’s social and geographical diversity, but also into its
diachrony and the impact of Tamil, the dominant language of the region. Even though, as
a creole, language contact between Portuguese and the languages of Sri Lanka (primarily
Tamil, in the case of the varieties studied here, but also potentially Sinhala, in earlier
stages) is part of the formative history of SLP, the contact dimension is especially relevant
to our synchronic analysis because, as explained in section 2, SLP is a minority nonofficial language and, as such, SLP speakers are at least bilingual.
The SLP speech community’s multilingualism is not alien to another fact which
informs our analysis of variation, viz. the language’s endangerment. Situations of
language obsolescence such as that observed in SLP are said to favor particular types of
structural change of the non-dominant languages, not because they are exclusive of
obsolescing languages, but because they are intensified or sped up in those contexts.
These include (to name but a few): overgeneralization of specific (marked or unmarked)
features, loss of phonological contrasts, morphological reduction, preference for analytic
over synthetic constructions, paradigm levelling, loss of allomorphy, loss of certain
grammatical categories, and stylistic shrinkage (see Campbell & Muntzel 1989; Sasse
2001; Palosaari & Campbell 2011; Aikhenvald 2020). In addition, it is also hypothesized
that obsolescence can make a receding language particularly permeable to the structural
influence of the community’s dominant language. Obsolescence may produce variability
at the individual level, through the flexibilization of earlier categorical rules (Sasse
2001:1671), or at the communal level, if its effects do not apply homogeneously across
3
the entire speech community. Given that the degree of SLP language loss is not socially
or geographically uniform, we will discuss whether (some of) the patterns of variation
identified may relate to language obsolescence – approached here in its dimension of
gradual communal (as opposed to individual) language loss.
The paper is organized as follows: section 2 provides necessary background
information concerning the object of research, the sources and methods of data collection,
and provides a brief sketch of past (19th-century) stages of SLP, collated from early
descriptive sources; section 3 describes the modern SLP personal pronoun paradigm,
contrasting previous descriptions with the scenario that emerges from an exploration of
the recently-collected corpus; section 4 discusses the observed instances of variation,
abstracting possible generalizations concerning the distribution of particular variants;
section 5 discusses, on the basis of the previous observations, the relevance of language
obsolescence to interpret certain observed patterns of variation; finally, in section 6, we
make some final remarks concerning the implications of this study for the description of
modern SLP and the practice of language documentation.
2. Background
The development of Sri Lanka Portuguese relates to Portuguese colonial presence on the
island, which lasted from 1505 to 1658. Over time, this creole acquired a relevant position
within that territory and across different communities, with an importance which justified
a flurry of 19th-century publications in or about this language, and Dalgado’s (1900:xxiixxiii) assertion that it was particularly prominent “entre os dialectos portugueses
coloniaes” (‘among the Portuguese colonial dialects’). However, the language
subsequently contracted, to the extent that it is now exclusively associated with one
subsection of Sri Lanka’s population: the Portuguese Burghers, i.e. Sri Lankans who
claim an Asian-Portuguese ancestry (see McGilvray 1982). By all accounts, SLP is
currently endangered (Nordhoff 2013; Pereira 2019; Cardoso et al. 2019).
At present, pockets of SLP speakers2 (see Fig. 1) are located in the Eastern
Province, in the cities of Trincomalee and Batticaloa (which host the largest
concentrations of speakers) and in nearby locations (e.g. Eravur, Valachchenai,
Kalmunai, and around Akkaraipattu) in the districts of Trincomalee and Batticaloa, but
also of Ampara. In addition to this, there is a small number of speakers in the Northern
Province, in the city of Jaffna.
It is important to mention that, while the communities of the Eastern Province are in close
contact with each other – with family relations extending across the region, considerable
2
Excluding those that resulted from recent migrations to other parts of Sri Lanka, especially the city of
Colombo.
4
mobility and opportunities for gatherings (such as weddings or community events), the
Jaffna community in the Northern Province has long been isolated from these dynamics,
so much so that the recent identification of speakers in that city came as a surprise to the
community in the Eastern Province. In addition, the process of language loss is especially
advanced in Jaffna, in which very few people – likely between 10 and 15 – speak SLP,
the youngest one (identified so far) being 62 years old.
Figure 1: Distribution of the SLP-speaking community in modern Sri Lanka.
When it comes to the Eastern Province, while the overall numbers are much larger,
language shift is also a real issue, and affects some regions and sections of the population
more than others. A recent sociolinguistic survey conducted among the Portuguese
Burgher community of the Eastern Province3 (between October 2017 and May 2018),
aimed at ascertaining fluency in SLP and language use among the community, gathered
information from 3094 respondents and provides some information in this respect.
Overall, 25.53% of respondents declared complete fluency or little difficulty in
using/understanding SLP, with an additional 15.68% declaring some difficulty in
speaking/understanding it. However, these values vary substantially across districts, with
a more robust knowledge of the language in Trincomalee (38.37% and 14.92%,
3
Jaffna could not be surveyed at the time, because the identification of SLP speakers in that city only
occurred towards the end of the research project.
5
respectively), a lower percentage of fluency in Batticaloa (25.56% and 19%, respectively)
and much smaller numbers in Ampara (15.60% and 8.07%, respectively) (see Cardoso et
al. 2019). Another important conclusion of the survey is that fluency in SLP is unevenly
distributed across age groups (see Pereira 2019), with a low percentage of respondents in
age groups below 40 declaring any knowledge of SLP at all (with values by decade and
across the 3 districts varying between 0% and 34.7%), a somewhat higher percentage in
age groups between 40 and 60 (with values varying between 22% and 45.3%) and a much
higher percentage among respondents over 60 (with values varying between 43% and
100%). This age distribution of fluency reveals the effects of a progressive break in the
transmission of SLP, which can be said to be undergoing a process of “gradual death”
(Campbell & Muntzel 1989:184-185), with the younger generations showing an ever
more evident shift to a different language – in this case, as also shown by the results of
the survey (Pereira 2019), mostly to Tamil.
Contrary to what happens with many other creoles, a significant number of written
sources for SLP have been produced, especially in the late 19th and early 20th century (see
Tomás 1992; Cardoso, Hagemeijer & Alexandre 2015; Smith 2016). Afterwards,
following a period in which the language received little attention (through most of the
20th century), the documentation and description of modern SLP resumed in the 1970s,
pushed forward in particular by Ian Smith’s research (e.g. Smith 1977, 1979a, 1979b,
2013). This work brought the language to the forefront of various debates within contact
linguistics, especially with reference to the mechanisms and speed of convergence (see
e.g. Smith 1979a; Bakker 2006) and resulted in a significant amount of grammatical
description.
Throughout this long history of SLP documentation, the various sources do record
linguistic diversity, but explicit explorations of variation are rare. Arguably, one reason
why community-internal variation has not become more prominent in SLP research is
that, until recently, the available spoken data (most prominently that included in Smith’s
1973 corpus) had been produced by relatively few speakers, mostly concentrated in the
city of Batticaloa.
However, the recent Documentation of Sri Lanka Portuguese project (henceforth
DSLP; see Cardoso 2017; Cardoso et al. 2019), hosted by the Centro de Linguística da
Universidade de Lisboa and funded by the Endangered Languages Documentation
Programme, has collected a more varied corpus of SLP speech, with language samples
from circa 150 speakers collected in nearly 50 different locations (towns or town areas)
scattered across 3 districts of Eastern Sri Lanka (Ampara, Batticaloa, and Trincomalee)
and 1 of Northern Sri Lanka (Jaffna). The DSLP corpus substantially expands the
diversity of sociolinguistic profiles for which spoken data are available, as well as the
geographical reach of the documentation. In fact, the inclusion of Jaffna in the
documentation is a complete innovation. The northern city was originally not
6
contemplated in the documentation project because the received wisdom was that, as in
many other former SLP-speaking strongholds within Sri Lanka, there were no speakers
of SLP left in Jaffna. However, a prospective visit to Jaffna in 2018 allowed the research
team to locate a few speakers and make the first few recordings of the northern variety.
Recently, in 2020, further fieldwork was conducted in Jaffna, resulting in an additional
18 hours of recorded speech (involving narratives, conversation, and elicitation sessions),
as a result of which this study is able to make the very first contribution to the description
of the Jaffna variety of SLP and to the exploration of the extent to which it coincides with
the eastern varieties. Unsurprisingly, given its breadth, the DSLP corpus reveals instances
of variation which had not been recorded before, including in the forms and uses of
personal pronouns – the focus of the present study.
2.1. Personal pronouns, diachrony and language contact
Cross-linguistically, personal pronouns constitute a universal or quasi-universal (Bhat
2004:30-31; Siewierska 2004:13) heterogeneous category comprising a small and closed
set of independent or bound forms whose primary function is to assign the deictic
category of grammatical person to the entities involved in a speech act, viz. the speaker,
the addressee, and other entities in the linguistic context. When dealing with personal
pronouns, there is however a wide range of linguistic issues to observe, since crosslinguistic data (see e.g. Helmbrecht 2013; Siewierska 2013) reveals that, in addition to
person (1st, 2nd, 3rd), they can also convey other distinctions, such as number (singular,
plural, dual, trial, etc.), gender (masculine, feminine, neutral) and, in some cases,
honorificity (degrees of politeness or formality) or clusivity (inclusive or exclusive
integration of the addressee). Like nouns, personal pronouns may also convey case
distinctions. Additionally, the pragmatic manipulation of a given distinction to convey a
non-original purpose (e.g. the well-known use of a second person plural pronoun for a
singular honorific referent) is also attested.
Personal pronoun paradigms are usually seen as three-person sets of forms, but
theoretical and typological studies (e.g. Benveniste 1971; Lyons 1977; Bhat 2004) based
on an analysis of the speech acts and the respective roles performed by personal pronouns
have pointed out the distinction between 1st and 2nd person pronouns, on the one hand,
and 3rd person pronouns, on the other. According to those studies, while 3rd person
pronouns are used for denoting anaphoric referents whose interpretation is dependent on
the linguistic context, 1st and 2nd person pronouns are deictic forms which denote the two
principal participant/speech roles of the clause in which they occur, viz. the speaker and
the addressee. This approach suggests that only 1st and 2nd person pronouns are personal
pronouns per se, assigning 3rd person pronouns to a different category – Bhat (2004:132),
7
for instance, following the analysis of a 255-language sample, considers 3rd person
pronouns to constitute an intermediate class between ‘personal pronouns’ and the socalled ‘proforms’ (i.e. demonstratives, indefinites and interrogative pronouns), arguing
for a particular similitude between 3rd person pronouns and demonstratives, demonstrated
by 49% of his language sample, in which third person pronouns morphologically overlap
or derive from demonstratives.
From a diachronic point of view, personal pronouns are “often considered to
belong to the most conservative part of grammar and [are] regarded as historically stable
closed-class items” (Ishiyama 2019:109), though not all forms are said to be equally
stable. A concomitant notion to diachronic stability is that it also manifests itself in
resistance to replacement by borrowing, which is why some (but not all) personal
pronouns normally feature in lists of basic vocabulary. If we compare Swadesh’s 100strong list of basic vocabulary with the more empirically-constructed Leipzig-Jakarta list
(see Tadmor 2009:68ff), we notice that both of them coincide in recognizing the status of
1SG and 2SG as stable items, whereas the Leipzig-Jakarta list – contra Swadesh –
ascertains the stability of 3SG and disconfirms that of 1PL; 2PL and 3PL pronouns are
absent from both lists. This makes the pronominal system an important testing ground for
diachronic hypotheses concerning e.g. historical relations between discrete language
varieties or particular instances of morphosyntactic change, and will aid us in interpreting
the variation we find within the modern SLP pronominal paradigm.
2.2. Early accounts of SLP personal pronouns
As mentioned above, SLP (earlier known as Ceylon Portuguese) has a relatively long
history of written documentation, due to its importance as a lingua franca during the
island’s colonial period. Dating back from the early 19th century, bibliographical sources
produced in or about SLP included grammars, vocabularies, phrasebooks, bilingual or
trilingual dictionaries, and translations of liturgical and biblical texts (e.g. Berrenger
1811; Callaway 1818, 1820, 1823; Fox 1819; Anon 1826; Newstead 1827, 1852, 1871;
Anon. 1851; Anon. 1863). As noticed by several authors and discussed in detail in Smith
(2016), the language in these sources is different from modern SLP in many fundamental
ways, and, since most of them were produced by non-native speakers, significant
interference and linguistic tweaking is expected to have taken place. Nevertheless, these
are the only available sources referring to a variety of SLP spoken throughout the 19th
century, therefore it is interesting to observe what they indicate in terms of the language’s
personal pronoun paradigm. In Tables 1 and 2, we have collated the nominative and nonnominative forms of personal pronouns found in some of these early written sources,
namely Berrenger (1811), Fox (1819), Callaway (1820) and Dalgado (1900):
8
Nominative
1SG
2SG
3SG.M
3SG.F
1PL
Berrenger
1811
eu
eu mesmo
voss/boss
voss mesmo
3PL.M
3PL.F
eu
eu mesmo
vos/vosso
vossamesmo
eli
eli mesmo
ela/elé
ela mesmo
nossé
nossé mesmo
vossé
vosse mesmo
elli
ellimesmo
ella
ellamesmo
nos
nossmesmo
vos, vosses
vossotros
mesmo
vossmesmo
elotros
elotros
mesmo
ellotros
ellotrosmesmo
2PL
3PL
Fox
1819
Callaway
1820
eu
Dalgado
1900
eu
vos/
tu/vós
vosse/vosses/vosse
merci
elle
elle
ella
ella
nos
nós/nossotros
volotros (to
inferiors)
vosotros (to
equals)
vosse-ellotros (to
superiors)
elles (of a few)
ellotros
vós/vossotros
ellotros
elles
ellas
Table 1: Nominative forms of personal pronouns in early (19th-c. and early 20th-c.) sources for
SLP.
This early SLP personal pronoun paradigm retains several forms of (16th-century)
European Portuguese. It is characterized by a three-person set of independent forms which
vary according to the pronoun’s grammatical function in an utterance. Accusative, dative,
genitive and ablative pronouns derive by and large from their nominative counterparts
(though sometimes also non-nominative forms, such as mi in the case of 1sG) and involve
recourse to certain prepositions (e.g. de voss ‘2SG.GEN’; per mi ‘1SG.DAT’) or,
exceptionally, the postposed case-marker -su (derived from Ptg. sua ‘3SG.GEN’).
Early SLP pronouns also encode a binary distinction of number (singular and
plural) and gender (masculine and feminine). In the case of gender distinctions, it is worth
noting that they only apply to 3rd person pronouns; all these sources record gender
distinctions in 3SG pronouns (along the lines of elli ‘1SG.M’ vs. ella ‘1SG.F’), but only
Dalgado (1900) registers their extension to 3PL pronouns.
9
Non-nominative
1SG.GEN
1SG.DAT
1SG.ACC
1SG.ABL
2SG
3SG.M
3SG.F
1PL
2PL
3PL
Berrenger 1811
de eu
ne eu
eu
de/por mi ~ pormi
de voss
ne voss
voss
de voss/por vos
de eli/d'eli
ne eli
eli
de/por eli
Fox 1819
de mi
per mi
mi
ne, de mi
de vos/vosse-su
per vos
vos
ne, de vos
de elli
per elli
elli
ne, de elli
d'ela
ne éla
ela
de/por ela
de nossé
ne nossé
nossé
de/por nossé
de vossé
ne vossé
vossé
de vosse/por vossé
de ella
per ella
-
ne, de ella
de nos
per/par nos
nos
ne, de nos
de vosotros-su
per vosotros
vosotros
ne, de vosotros
de ellotros
per ellotros
ellotros
ne, de ellotros
Table 2. Non-nominative forms of personal pronouns in early (19th-c. and early 20th-c.) sources
for SLP.
Berrenger (1811) and Fox (1819) include in their paradigms the combination of the
nominative forms of personal pronouns with the element mesmo, as in ela mesmo
‘herself’, suggesting the existence of a reflexive strategy constructed this way.
Finally, these sources also record politeness distinctions, with certain 2nd person
pronouns attributed to particular addressees in accordance with their social status, as
explicitly indicated by Callaway (1820), who explained that volotros was used for
“inferiors”, vossotros for “equals”, and vosse-ellotros for “superiors”. Another instance
of honorificity, clearly reminiscent of Portuguese, involves the opposition between the
10
familiar t-form tu ‘2SG’, when addressing one individual, and the v-form vos/vós
‘2SG/2PL’, used by default for more than one entity, but also as a polite form for single
individuals; yet, this politeness distinction is only given in Dalgado (1900). In addition,
Callaway (1820) includes a noun-based form of polite address, vosse merci (< Ptg. vossa
mercê ‘your mercy’).
2.3. Data and methodology
The DSLP corpus of modern SLP (Cardoso 2017) which provides the primary data for
our synchronic study consists of over 49 hours of video and/or audio recordings of
naturally occurring speech and musical performances, produced between 2015 and 2020
in several field trips to various locations in the Ampara, Batticaloa, Jaffna and
Trincomalee districts of Eastern and Northern Sri Lanka (Cardoso 2017).4 Interviews
involved single individuals, small groups (e.g. couples, households) or large groups (e.g.
extended families, gatherings). The consultants were often invited to address topics
related, but not restricted, to their personal experience or to Burgher identity (language,
music, cuisine, crafts, occupations, religion, daily life), but there were also sessions of
lexical and grammatical elicitation.
Song lyrics constitute an important part of the corpus, because the DSLP project
combined linguistic and ethnomusicological research. However, there is good reason to
suspect song lyrics to be particularly conservative and to potentially include linguistic
resources which are learnt by rote, which, though interesting and diachronically revealing,
should be approached as a separate corpus. Therefore, we have excluded the recordings
of songs from our study and focused on naturally occurring speech.
While there is considerable sociolinguistic diversity among the 152 recorded
consultants, some profiles (geographical and social) are especially well represented.
Table 3 below provides some basic information on the profile of speakers recorded in the
DSLP corpus:
Eastern Province
Sex
F
M
Northern
Province
Batticaloa
Kalmunai
Eravur
Valaichchenai
Trincomalee
Jaffna
43
39
3
3
2
-
6
5
26
20
1
4
4
More detailed information on the corpus (constitution, transcription conventions, workflow and types of
archived materials) is given in Cardoso et al. (2019). The materials are made available on the Endangered
Languages Archive, at https://www.elararchive.org.
11
Age
0-14
15-24
25-64
≥ 65
N/A
4
3
16
9
50
2
1
3
2
4
7
1
4
21
13
7
2
3
-
Total
82
6
2
11
46
5
Table 3. Distribution of speakers in the DSLP corpus (Cardoso 2017) by region, sex, and age
group (N=152).
As can be seen in Table 3, the gender distribution only slightly favors women, with 81
female consultants (53%) and 71 male consultants (47%). Consultants vary in age, despite
the fact that there is a preponderance of middle-aged or elderly interviewees, as a result
of the current sociolinguistic distribution of SLP fluency described above. However, there
is a clear predominance of speakers from the urban areas of Batticaloa and, to a lesser
extent, Trincomalee. The towns of Eravur and Valaichchenai also belong to the Batticaloa
district, whereas Kalmunai is the only town standing for the Ampara district. The number
of consultants from Jaffna may seem modest but, according to our current knowledge, it
represents a very large percentage of all the speakers in the Northern Province.
The data for this study was gathered by searching the DSLP corpus within
multiple ELAN (Brugman & Russel 2004) annotation files, automatically extracting all
personal pronouns (and their phonetic and morphological variants) produced by each
speaker. The collected sample was then organized in a spreadsheet, filtering out target
word forms that occurred in song lyrics and those that had been produced by the
interviewers.
3. Modern SLP personal pronouns
Within the modern SLP nominal domain, case distinctions involve the addition of a series
of postposed morphemes – which Smith treats as suffixes in the case of the oblique (i.e.
accusative + dative), genitive, and locative, and as postpositions in other cases – to the
nominative (or, in the case of some postpositions, the genitive) form of the noun (see
Smith 2013:113). Yet another productive nominal suffix is the plural marker -s. The
addition of these morphemes to nouns produces transparent multimorphemic forms
containing case and number information (e.g. luváára ‘place’ ~ luvááras ‘places’ ~
luváárantu ‘in (the/a) place’).
12
When it comes to the personal pronoun paradigm, however, the situation is
somewhat different. While the same additional morphemes also occur, they do not
account for all pronominal forms, several of which are monomorphemic and suppletive.
Such forms occur in the nominative, oblique, and genitive series, which will constitute
the core of our analysis in this study.
Ian Smith’s descriptive work on SLP (Smith 1977; 2013) records the pronominal
forms given in Table 4:
1SG
2SG.NHON
3SG.M.NHON
3SG.F.NHON
3SG.HON
1PL
2PL
3PL.M.NHON
3PL.F.NHON
3PL.HON
Nominative
eev
boos
eli
ɛla
osiir
noos
botus
elis
ɛlas
etus
Oblique
parim/parmi
boos-pa
eli-pa
ɛla-pa
osiir-pa
noos-pa
botus-pa
elis-pa
ɛlas-pa
etus-pa
Genitive
miɲa
bosa
eli-su
ɛla-su
osiir-su
nosa
botus-su
elis-su
ɛlas-su
etus-su
Table 4. SLP personal paradigm (adapted from Smith 2013:114).
As can be seen in Table 4, honorificity is an important sociolinguistic factor encapsulated
in the pronominal paradigm. In 3rd person, honorific and non-honorific address involve
dedicated forms. In the case of the 2nd person, however, no distinction is encoded in plural
pronominal forms, but the 2PL form botus is also used as a honorific form of address for
2SG (Smith 1977:69), thereby creating a functional syncretism between 2SG.HON and 2PL.
In addition, non-honorific 2nd person pronouns may also be avoided by substituting them
with a title such as sinhoor ‘gentleman, mister’ (Smith 2013).
In this paradigm, we can also observe that the transparent use of plural suffix -s is
restricted to non-honorific 3M and 3F pronouns, whereas all other plural pronominal forms
are suppletive. When it comes to case distinctions, the oblique suffix -pa is recognized in
all oblique forms except 1SG, while the genitive suffix -su is not applied in any 1st person
pronoun nor in non-honorific 2SG.
Smith’s description of the SLP pronominal paradigm, given in Table 4, is based
on a corpus (Smith 1973) containing approximately 15h of interviews, collected mostly
in Batticaloa. With nearly 50h and 152 interviewees, the DSLP corpus is larger and was
collected in several locations of the Batticaloa, Trincomalee, Ampara, and Jaffna districts
(see section 2.3). Unsurprisingly, this corpus provides alternative forms for several
pronouns, resulting in the updated paradigm given in Table 5. Orthography varies slightly
with respect to Table 4, reflecting the DSLP team’s preferred orthography (for an
13
explanation of which, see Cardoso et al. 2019:12-16); all forms undocumented in previous
accounts of the modern SLP paradigm are underlined:
1SG
Nominative
Forms Tokens
eev
2203
ee
77
2SG.NHON
3SG.M.NHON
3SG.F.NHON
3SG.HON
boos
eli
éla
osiir
esiir
497
167
100
815
24
1PL
2PL
nóós
botus
botrus
elis
élas
etus
etrus
2599
377
58
31
3
1139
333
3PL.M.NHON
3PL.F.NHON
3PL.HON
Oblique
Forms
Tokens
páármi
337
paarmi
62
páámi
14
paami
204
pááim
26
paaim
45
páármi-pa
3
paarmi-pa
3
paami-pa
29
paaim-pa
104
eev-pa
56
boos-pa
241
eli-pa
57
éla-pa
34
osiir-pa
193
nóós-pa
botus-pa
botrus-pa
elis-pa
élas-pa
etus-pa
etrus-pa
985
113
6
13
1
384
86
Genitive
Forms
Tokens
minha
1409
eev-su
22
paami-su
16
minha-su
2
eev-pa
1
bósa
eli-su
éla-su
osiir-su
343
15
22
144
nósa
botus-su
botrus-su
[elis-su]
[élas-su]
etus-su
etrus-su
2182
87
11
0
0
259
44
Table 5. SLP personal pronouns recorded by the DSLP project, with number of tokens in the
DSLP spoken corpus (Cardoso 2017); forms in square brackets were attested in elicitation and
fieldnotes but do not occur in the oral corpus.
The nature of the variation encapsulated in Table 5 is not uniform. In some cases, what is
at stake are mere differences in the form of particular suppletive pronouns, involving
alternations at the level of certain segments (as in 3SG.HON, osiir ~ esiir), the
addition/deletion of certain segments (as in 2PL and 3PL, botus ~ botrus, etus ~ etrus, and
derived forms thereof), or metathesis (as in 1SG.OBL, páámi/paami ~ pááim/paaim).
14
In other cases, variation involves oppositions between monomorphemic5 (i.e.
suppletive) and bimorphemic (i.e. analytic) forms for oblique and genitive pronouns. In
general, across the paradigm, bimorphemic forms consist of the addition of the
corresponding case morpheme (-pa for oblique and -su for genitive) to a base which is
equal to the nominative pronoun; as such, and considering that this also holds true of the
use of these case-markers with nouns, we can say that it constitutes the regular pattern.
However, in those domains in which we observe variation, namely 1SG.OBL and 1SG.GEN,
some bimorphemic alternatives follow this pattern (the cases of eev-pa and eev-su), while
others do not (i.e. they select a base form equivalent to a monomorphemic oblique or
genitive pronoun); and, as for the latter, either the case associated with the base form and
the case-marker coincide (as in páármi/paarmi/paami/paaim-pa, and minha-su) or they
diverge (as in paami-su). The one form which falls outside of these possibilities is the
single recorded instance of eev-pa used in a clearly genitive context, in which the oblique
morpheme intrudes in the distributional space of the genitive; we interpret this as a matter
of non-canonical case-selection and, therefore, will treat the occurrence as an outlier.
In addition to this – though not evident from Table 5 –, other instances of variation
in pronominal selection can be observed in the domain of pragmatics, involving the
system of honorific address (in 2SG and 3SG/3PL), and a particular case of number
syncretism pertaining to 1st person genitive. These cases of variation in the DSLP corpus
will be discussed in the next section.
4. Variation
4.1 Variable forms
When we look at the modern SLP pronominal paradigm (Tables 4 and 5), we identify
different pronominal forms for 1SG, 2PL, 3SG.HON, and 3PL.HON. Some of the oppositions
involve morphological differences and will be discussed in 4.2; others, however, are of a
different nature. Some are purely phonetic:
- the difference in vowel height between páármi/páámi/pááim (with a long low central
vowel [a:]) and paarmi/paami/paaim (with a long near-low central vowel [ɐ:]);
5
The question may arise why 1SG oblique forms páármi/paarmi/páámi/paami/pááim/paaim are considered
monomorphemic in modern SLP, especially when early sources record what appears to be a preposition
par/per/por with various personal pronouns (see Tables 1 and 2) and nouns, including the 1SG sequences
por mi and per mi. That early preposition but also the modern oblique case suffix -pa are indeed derived
from the Portuguese preposition para ‘for’ or por ‘by’. But, while analytical par/per/por mi does appear to
be the diachronic source of páármi/paarmi/páámi/paami/pááim/paaim, we cannot recognise a preposition
or prefix here because it occurs nowhere else in the modern SLP nominal system (i.e. it is not productive);
a prefix pa- does exist, but it is strictly a verbal prefix marking the infinitive (see Smith 2013).
15
- the difference in vowel backness and roundedness between osiir (the most frequent
variant, with a high-mid back rounded vowel [o]) and esiir (with a high-mid front
unrounded vowel [e]).
Some other differences are the result of metathesis:
- the difference between páámi/paami and pááim/paaim (in the DSLP corpus);
- the difference between parmi and parim (recorded only in Smith’s corpus, see Table 4).
In other cases still, observed differences involve addition/deletion of a segment:
- the difference between eev and ee (the latter only attested in the Jaffna variety);
- the difference between páármi/paarmi and páámi/paami;
- the difference between botus and botrus;
- the difference between etus and etrus.
With the exception of the eev ~ ee opposition, all cases of deletion refer to the flap,
creating an opposition between forms with and without [ɾ]. For ease of reference, we will
identify one type as R+ (i.e. with [ɾ]) and the other as R- (i.e. without [ɾ]). Table 6
indicates the percentage of speakers in the DSLP corpus who use 1SG.OBL, 2PL and
3PL.HON pronouns of the R+ type, of the R- type, or alternate between both:
R-
Both R- and R+
R+
Total
1SG.OBL
19.5% (17)
34.5% (30)
46% (40)
100% (87)
2PL
84.4% (65)
11.7% (9)
3.9% (3)
100% (77)
3PL.HON
77.4% (89)
20.9% (24)
1.7% (2)
100% (115)
Table 6: Use of competing 1SG.OBL, 2PL and 3PL.HON forms (% speakers + number of speakers).
While, in all cases, there is a considerable percentage of speakers who alternate between
competing forms, the highest proportion is always found among those for whom one of
the alternatives is categorical. However, the preference is not always the same. In the case
of 1SG.OBL, R+ forms are synchronically dominant, with 46% of speakers using those
consistently, and an additional 34.5% alternating between R+ and R- forms. In 2PL and
3PL.HON, on the other hand, the preference goes to R- forms and is even more pronounced,
with a vast majority of speakers using only R- forms (84.4% and 77.4%, respectively),
and very few only R+ forms (3.9% and 1.7%, respectively). This particular clustering is
interesting in that it coincides with a difference in the phonological context involved in
16
the R+ ~ R- alternation. Even though all cases involve the presence or absence of a flap,
in 1SG.OBL, the segment (when realized) occurs in a syllable coda, whereas, in 2PL and
3PL.HON, what is at stake is an alternation between a simple onset [t] and a complex onset
[tɾ] in the pronoun’s final syllable.
A closer look at the distribution of R+ and R- forms reveals that there is a single
regional variant in which the competition we just described does not apply. The data
collected in Jaffna contains no instances of R+ forms, not just for 2PL and 3PL.HON
pronoun, but also for 1SG.OBL. Elsewhere, the synchronic instability in the pronunciation
of these pronominal forms raises the question of whether we are witnessing a process of
diachronic change – and, if so, in which direction it proceeds. We would like to argue for
a scenario which interprets R+ forms as older or more conservative, and R- forms as more
recent. Bearing in mind the distribution in Table 6, this hypothesis implies that the process
of change, overall, is less advanced in 1SG.OBL, and approaching its completion in 2PL
and 3PL.HON; and completed in Jaffna.
The rationale behind this proposal rests on a few considerations. First of all, the
etymological criterion favors the early presence of [tɾ] in all these cases: the
páá(r)/paa(r)- element of 1SG.OBL derives from the Ptg. prepositions para ‘for’ or por
‘by’ (see Note 5), and the -t(r)us of 2PL and 3PL.HON derives from the Ptg. pronoun outros
‘others’. As a matter of fact, the corresponding forms in 19th-century sources (see Tables
1 and 2) all reflect the etymological flap. It is true, as already mentioned, that the data in
these sources needs to be taken with a grain of salt, but it is significant that all sources
surveyed coincide in this respect.
Another fact which reinforces the likelihood of SLP having preserved the flap in
the case of 2PL and 3PL.HON (rather than introducing it at a later stage) is that, across its
lexicon, we observe a general tendency for similar etymological complex onsets
combining a plosive and a flap to have been preserved, and they are to be found in wordinitial and word-medial position alike (see Smith 1977:43-52). A few examples include
the modern SLP words létriiya ‘string hopper’ (from Ptg. aletria ‘vermicelli pasta’),
káátru ‘four’ (from Ptg. quatro ‘four’), lembráá ‘to think’ (from Ptg. lembrar ‘to
remember’), triiya ‘to bring’ (from Ptg. trazer ‘to bring’), trukáá ‘to (ex)change’ (from
Ptg. trocar ‘to (ex)change’), among others. In fact, SLP’s acceptance of this type of
complex onset has even made it possible for the language to develop a few which were
not present in the Portuguese etyma, either as a result of vowel deletion (often involving
other segmental transformations), as in kambráám ‘shrimp’ (from Ptg. camarão ‘shrimp’)
and páástru ‘bird’ (from Ptg. pássaro ‘bird’), of metathesis, as in brumeey ‘red’ (from
Ptg. vermelho ‘red), or of other types of innovations, as in usprutáál ‘hospital’ (from Ptg.
hospital).
According to this interpretation, then, the R- 2PL and 3PL.HON pronominal forms
constitute exceptional cases and the resulting simple ~ complex onset alternation reflects
17
diachronic innovation. However, they are not alone, as there is at least one other case of
alternation between simple and complex onsets involving a plosive and a flap, viz.
primeer/prumeer ~ pimeer/pumeer ‘first, earlier’. Considering that this term derives from
Ptg. primeiro ‘first, earlier’, which contains the complex onset in question, it would also
be difficult to argue for the loss and later reintroduction of the flap. As an important
adverbial, primeer ~ pimeer is highly frequent in discourse, just as botrus ~ botus and
etrus ~ etus, which may be at the root of the motivation for onset simplification, in all
three cases. In addition, the R- form pimeer is much more frequent in the global DSLP
corpus (525 tokens) than any of the competing forms (the closest being primeer, with 191
tokens), a synchronic predominance of the R- form which mirrors what we described for
the 2PL and 3PL.HON pronouns; and, consistently with our observations with respect to
pronominal forms, R+ primeer/prumeer is entirely absent from the Jaffna data.
4.2 Monomorphemic vs. bimorphemic forms
In the DSLP corpus, we find competing monomorphemic (suppletive) and bimorphemic
(analytic) forms for the oblique and genitive cases of 1SG (see section 3).
For 1SG.OBL, Smith records only monomorphemic parmi or parim (see Table 4).
As mentioned earlier, the DSLP corpus also includes the suppletive oblique form
páármi/paarmi, as exemplified in (1), as well as variants páámi/paami and pááim/paaim
(Table 5):
(1)
páármi
káátru podhiyáás,
doos mááchi, doos féémia.
1SG.OBL
four children
two
male
two female
‘I have four children, two sons and two daughters’. (Trincomalee; Cardoso
2017:009_1)
However, a few bimorphemic variants are also to be found: a) eev-pa, as in (2); and b) a
set of forms, as in (3), consisting of a variant of the suppletive oblique pronoun plus the
case-marker, viz. páármi/paarmi-pa, paaim-pa, and paaim-pa.
(2)
eev-pa
triinta sees "years" ta-fikáá.
1SG.NOM-OBL thirty six
years PRS-become
‘I am turning thirty-six years (of age)’. (Trincomalee; Cardoso 2017:slp071_1)
(3)
paaim-pa
ya-ka-iskisa
1SG.OBL-OBL
PST-PFV-forget
‘I have forgotten’. (Jaffna; Cardoso 2017:slp069_4)
18
Given that, in the oblique series, all forms but those of 1SG are strictly analytic (consisting
of the application of the oblique case-marker -pa), the formation of bimorphemic 1SG
forms can be seen as a case of regularization of the paradigm. However, there is a
difference between eev-pa and the other bimorphemic forms, in that the former constitutes
the full regularization of the oblique pronoun formation strategy by which the oblique
case-marker selects a nominative base form of the pronoun, whereas the latter adds that
marker to an intrinsically oblique pronominal form, thereby resulting in redundant double
case-marking.
The latter strategy (i.e. the selection of a non-nominative base) may be explained
by transfer from Tamil. In this language, pronouns (as well as nouns, see Schiffman
1999:27) have two monomorphemic forms, the nominative and the oblique, and casemarkers attach to the oblique base. In the case of personal pronouns (see Schiffman
1999:117-119), distinct nominative and oblique forms exist for 1st and 2nd person forms
(e.g. naan ‘1SG.NOM’ ~ en ‘1SG.OBL’; nii ‘2SG.NOM’ ~ on ‘2SG.OBL’), but not for 3rd
person (e.g. avan ‘3M.SG.NOM/OBL’); and dative forms of the pronoun involve the
addition of the dative case-marker -(u)kku to the oblique base, whenever the distinction
holds (e.g. enakku ‘1SG.DAT’; onakku ‘2SG.DAT’), or to the invariable base, in the other
cases (e.g. avanukku ‘3M.SG.DAT’). This pattern may explain the option of certain SLPspeakers to select a non-nominative base for the oblique suffix in 1SG.OBL, but it should
be noted that these SLP forms and the Tamil suffixed personal pronouns are not entirely
equivalent. The Tamil oblique actually carries genitive semantics, since the bare oblique
form (of a noun or pronoun) can indicate possession, even though there are also genitive
markers; and, in SLP, the oblique conflates dative and accusative case. Therefore, the full
application of the Tamil model to the formation of SLP bimorphemic pronouns should
result in the selection of a genitive base (which does occur, though marginally, in the
genitive form minha-su; see below).
With 139 occurrences produced by 11 speakers, double case-marked forms
constitute a substantial proportion of 1SG.OBL forms in the corpus. However, they are
clearly associated with the Jaffna variety: 4 out of 5 speakers interviewed in Jaffna
produce double case-marked forms, and their speech accounts for 130 tokens, which
represents 81.8% of the 159 instances of 1SG.OBL forms collected in the Jaffna section of
the corpus; and globally, while the 4 Jaffna speakers make up just 36.4% of the 11
speakers who use such forms, they produce 93.5% of all occurrences of double casemarked 1SG.OBL pronouns in the DSLP corpus.
Having established the almost exclusive association of double case-marked
1SG.OBL forms with the Jaffna variety of SLP, let us now consider the use of the remaining
forms. Oblique eev-pa, the bimorphemic form which represents the simple regularization
of the paradigm with respect to 1SG.OBL, occurs 56 times in the DSLP corpus. Table 7
19
indicates the percentage of speakers who use this 1SG.OBL form only, suppletive forms
only, or alternate between both:
Eev-pa only
Both
Suppletive only
Total
6.7% (7)
16.3% (17)
77% (80)
100% (104)
Table 7: Use of eev-pa or a suppletive 1SG oblique form (% speakers + number of speakers).
The distribution in Table 7 clarifies that the suppletive strategy is clearly dominant for
1SG.OBL. Nonetheless, nearly a quarter of all speakers in the DSLP corpus employ eevpa, although most of those alternate between that bimorphemic form and a
monomorphemic one. Looking at the profile of these speakers who produce eev-pa, it is
difficult to assign the overall distribution of this form to particular age, gender, or
geographical groups. The only possible generalization is that all speakers for whom the
use of the bimorphemic eev-pa 1SG.OBL pronoun is categorical are under 40 years of age.
From an “apparent time” perspective, this distribution is consistent with a scenario of
language change in progress, but we believe the particular directionality of change has a
deeper meaning (see section 5).
When it comes to 1SG.GEN pronominal forms, the DSLP corpus records a
suppletive form minha (equivalent to Smith’s miɲa), exemplified in (4):
(4)
minha
pááy rábaana
lo-dááy.
1SG.GEN
father rabana
HAB-play
‘My father used to play the rabana6’. (Batticaloa; Cardoso 2017:slp040_3)
In addition, similarly to the oblique forms, the corpus also contains three bimorphemic
forms (plus a single occurrence of eev-pa in a genitive context, which, as explained in
section 3, we interpret as an instance of non-canonical case-selection, and therefore
disregard here): a) eev-su, as in (5), in which the regular genitive case-marker -su attaches
to a nominative pronominal base; b) paami-su, as in (6), in which the case marker selects
an oblique base form of the pronoun; and c) minha-su, as in (7), in which the genitive
case-marker attaches to the suppletive genitive pronoun; however, this form – which
constitutes yet another case of redundant double case-marking and, as explained above,
closely matches the Tamil model for the formation of non-nominative pronouns – is only
produced twice, by a single consultant from Jaffna.
6
A rabana is a type of traditional drum used in Sri Lanka, including in the musical traditions of the
Portuguese Burghers.
20
(5)
eev-su
kázmeentu "two thousand nine, August second"
ya-macháá.
1SG.NOM-GEN wedding two thousand nine, august second
PST-happen
‘My wedding took place on August 2nd, 2009.’ (Batticaloa; Cardoso
2017:slp020_6)
(6)
paami-su
kriyaansa
ya-ka-kazáá.
1SG.OBL-GEN children
PST-PFV- marry
‘My children have gotten married.’ (Jaffna; Cardoso 2017:slp064_1)
(7)
minha-su
santáá fáátu.
1SG.GEN-GEN sit
thing
'[This is] my chair.' (Jaffna; Cardoso 2017:slp2010_1)
The form paami-su is somewhat parallel to the oblique forms páármi/paarmi-pa, paamipa, and paaim-pa discussed above in that a case-marker selects an oblique base form of
the personal pronoun. In this case, however, it does not result in redundant double casemarking. In fact, with only 16 occurrences in the DSLP corpus, paami-su is also rather
marginal. Nonetheless, the distribution of these few cases is relevant, because they are all
produced by 3 speakers from Jaffna who are part of the group of 4 whose speech
accounted for almost all instances of double case-marked 1SG.OBL forms (see above).
This reveals a certain consistency in the grammar of these speakers from Jaffna and, at
the same time, reinforces the link between non-nominative base selection in case-marked
1SG pronouns and the Jaffna community.
Genitive eev-su constitutes a more canonical case of paradigm regularization, not
only because it mirrors the bimorphemic constitution of most other genitive pronominal
forms, but also in that it makes the genitive case-marker attach to the expected nominative
base form of the pronoun. With only 22 occurrences in the corpus, it is less marginal than
paami-su, but also not particularly frequent. Table 8 below indicates the proportion of
speakers who make categorical use of this form, of the suppletive form minha, or alternate
between both:
Eev-su only
Both
Minha only
Total
4,2% (5)
3,3% (4)
92,5% (112)
100% (121)
Table 8: Use of eev-su or the suppletive 1sg genitive form minha (% speakers + number of
speakers).
When we compare Tables 7 and 8, we notice that, in the case of the genitive, the use of
the bimorphemic form – both categorically and variably – is less prevalent than in the
21
oblique case. Looking at the sociolinguistic profile of the speakers, it is also more difficult
to associate this strategy to a particular age, sex, or geographical provenance, but it is
noteworthy that 4 out of the 9 speakers who produce this form are under 40 – 2 of whom,
in fact, under 16 years of age. In addition, another fact also reinforces the connection
between bimorphemic 1SG pronouns and lower age groups: only 2 consultants (both from
Batticaloa) make exclusive use of both eev-pa for 1SG.OBL and eev-su for 1SG.GEN, one
aged 12, and the other in her late 20s.
4.3 Honorificity
As we noticed in Tables 4 and 5 above, the full system of honorificity encoded in the
pronominal paradigm of modern SLP includes different forms (HON = ‘honorific’ and
NHON = ‘non-honorific’) for 2nd and 3rd person, which we will explore here in turn. In
terms of their distribution in discourse, by and large, NHON pronominal forms are
commonly associated with referents younger than the speaker – especially if the referents
are children or teenagers – and HON pronominal forms are used in all other cases.
In 2nd person, there is a 2SG.NHON form boos (with a suppletive genitive form,
bósa) used for familiar addressees (children especially), and a 2PL form botus or botrus
(see 4.1) which, in addition, functions as a 2SG.HON form of address, creating a functional
syncretism between 2PL and 2SG.HON, as previously described by Smith (1977:69).
Therefore, in reality, honorificity distinctions in 2nd person only surface in the singular.
Examples in (8) and (9) demonstrate the use of both forms with singular addressees:
(8)
eev kum dáádha kum ta-vii
boos
mee
nun-teem
áki.
1SG and father and PRS-come
2SG.NHON
FOC
NEG-be
here
‘Me and my father came, but you [= familiar addressee] weren't here.’
(Trincomalee; Cardoso 2017:slp077_1)
(9)
mesa-papiyáá botus.
OBLG-speak 2SG.HON
‘You [= respected single addressee] must speak.’ (Kalmunai; Cardoso
2017:slp027_1)
As demonstrated in Table 5, 2SG.NHON forms (boos, boos-pa and bósa) alone constitute
the majority of 2nd person pronouns in the DSLP corpus. In addition, many occurrences
of botus, botus-pa and botus-su also have a singular reference, and are therefore 2SG.HON
pronominal forms. The fact that 2SG pronouns outweigh 2PL pronouns in the DSLP corpus
is perhaps not surprising and derives from its very nature, considering that it contains a
22
large number of recording sessions involving a single interviewer and/or a single
interviewee.
In 3rd person, as shown in Table 5, NHON pronouns code gender distinctions in
both singular and plural, and their plural forms are transparently constructed with the
plural suffix -s. HON pronouns, on the other hand, have suppletive forms for singular and
plural and do not encode gender. Examples (10) and (11) demonstrate the use of HON and
NHON 3SG pronouns:
(10)
"portuguese" podhi unga
ya-tinha,
páármi
eli-pa
portuguese
boy one
PST-PST.be
1SG.OBL 3SG.NHON-OBL
nuku-sava.
NEG-know
‘There was a Portuguese boy, I didn't know him.’ (Batticaloa; Cardoso
2017:slp037_1)
(11)
avóóra osiir-pa
indung siinku áánu teen osiir-su
prenda
now 3SG.HON-OBL still
five year EXS 3SG.HON-GEN studies
kaváá-pa.
finish-PURP
‘He still has another five years to go to finish his studies’ (Trincomalee; Cardoso
2017:slp005_1)
Table 9 below shows the distribution of HON and NHON 3rd person pronouns in the corpus,
revealing, first of all, that NHON pronouns are much less common than HON pronouns:
NHON
HON
SG
PL
Total
89,2% (395)
10,8% (48)
100% (443)
34,5 % (1176)
65,6% (2245)
100% (3421)
Table 9: Distribution (number of occurrences) of 3rd person HON and NHON pronouns in the
DSLP corpus (Cardoso 2017).
Table 9 also makes it clear than NHON pronouns are especially infrequent with plural
reference. In this respect, it is interesting that, in addition to the canonical contexts in
which we would expect the selection of a NHON pronominal form (i.e. for younger or
familiar referents), in the case of 3PL we also see NHON pronouns applied to generic
referents, as demonstrated in example (12):
23
(12)
nóós
1PL
isi
nuku-oyáá,
NEG-see
faláátu.
DEM
QUOT
elis
3PL.M.NHON
ta-faláá ‘Malaysia’-su
PRS-say Malaysia-GEN
“wood”
wood
‘We didn't see it, they [i.e. people] say it's Malaysian wood.’ (Batticaloa; Cardoso
2017:slp021_2)
In the DSLP corpus, the application of pronominal honorificity distinctions functions
largely as expected, with one notable exception: the speech of the speakers from Jaffna.
The system of honorificity in Jaffna appears to be impoverished in comparison with the
speech of speakers from other regions represented in the corpus. In the case of the 2nd
person, this is manifested in the fact that Jaffna speakers appear to use mostly boos, boospa and bósa (329 occurrences vs. 26 in which the honorific form is used) in contexts
where other varieties would select a HON form, thereby contradicting the robust functional
syncretism between 2PL and 2SG.HON observed in the other varieties. When it comes to
the 3rd person, politeness oppositions also appear to be non-existent, as one singular
option and one plural option are never recorded in the available data: in the singular,
M.NHON eli and F.NHON éla occur, but the HON form osiir does not; whereas, in the plural,
HON etus is attested, whereas NHON elis or élas are not.
Another instance of variation which needs to be mentioned here involves a partial
syncretism between 1SG.GEN and 1PL.GEN pronouns, which has not been previously
described in the literature. While, as indicated in Tables 4 and 5, minha is the canonical
1SG.GEN pronoun (with the alternative forms discussed in 4.2 above) and nósa is the
canonical 1PL.GEN pronoun, the DSLP corpus contains several instances of nósa used in
contexts that are clearly singular, such as in example (13) – in which, in a strictly
monogamous community, the referent of the pronoun (a spouse) can only be a single
person; contrast it with example (4) above.
(13)
nósa
máriidu ja-ka-mura-pa
dispoos, eev kustuura
1PL.GEN
husband PST-PFV-die-OBL
after
1SG sewing
mee ya-kusa.
FOC
PST-sew
‘After my husband died, I started sewing.’ (Trincomalee; Cardoso 2017:slp003_5)
A study of such cases in the DSLP corpus reveals that this type of variation in genitive
1SG pronominal reference bears some relationship with notions of honorificity, in that this
use of nósa appears to be restricted to contexts in which speakers refer to their rapport
with a relative who commands a degree of honorificity. In the corpus, we only identify
24
this specific use of nósa in sentences that refer to a husband (as in (13)), a wife, an uncle,
or adult siblings.
5. Discussion
While, as discussed in section 1, researchers dealing with variation in an endangered or
lesser-studied language often face the challenge of having no prior knowledge of it, in
our case, Smith’s (2013) description of the SLP personal pronoun paradigm assisted us
in identifying variation in this domain. In general, our study confirms that the pronominal
forms previously described in the literature – i.e. those in Table 4 – are indeed the most
robust and most frequent in discourse, and for that reason can be considered canonical,
the only exception being the alternate 1SG.OBL form parim reported by Smith (2013:114),
but absent from the DSLP corpus.
Having said that, our survey also unearthed numerous alternate forms and
instances of variability in the expression of personal pronouns. In some cases, the
available data does not reveal a clear association between a variant and a given group of
speakers. In other cases, however, a distribution pattern emerges, associating a variant
with a subset of the speech community, either socially or geographically defined. In most
of them, the fact that these groups of speakers are those among which language shift is
most prevalent intersects with the fact that the variant in question is consistent with one
of the processes of structural change expected to result from language loss (see section 1)
to suggest that language obsolescence does underlie some of variability observed in
modern SLP.7
As described in section 2, the youngest sections of the Burgher community report
the lowest levels of proficiency in SLP and frequency of use of the language (Pereira
2019). Two alternate pronominal forms are especially associated with this sociolinguistic
profile: the bimorphemic 1SG.OBL eev-pa and the 1SG.GEN eev-su (see 4.2). The
association is especially strong in the case of eev-pa, since it was observed that all
speakers for whom this form is categorical (i.e. who never use one of the suppletive
forms) are under 40; in the case of eev-su, the association is based on the fact that, of the
few speakers who use it at all (whether or not in competition with the suppletive
alternative minha), nearly half are under 40; in addition, there are only two consultants
7
There are two cases in which a particular variant has a clear geographical association and yet it is not so
clear that it necessarily derives from language loss: a) the fact that the data from Jaffna contains no R+
pronominal forms (while R+ and R- compete elsewhere) does result in a reduced level of societal variation
in this variety and in this domain, but whether this development coincides with any of the processes
impacted by obsolescence cannot be ascertained without a more detailed study of Jaffna SLP phonology;
and the same could be said of b) the fact that the reduced form of 1SG ee is only identified in speech from
Jaffna.
25
for whom both eev-pa and eev-su are categorical, and they belong to this age bracket.
What unites these two forms, within the framework of the variation described in section
4, is that they represent instances of regularisation by analogy with the structure of nearly
all other non-nominative forms in the pronominal paradigm. If we consider suppletive
1SG forms to be older than the suffixed 1SG forms of modern SLP (as suggested by the
early data in Tables 1 and 2, and by proposed scenarios of convergence towards Tamil
affecting word order; see Bakker 2006), we must admit that the preference for eev-pa and
eev-su constitute innovations. So, these speakers engage in paradigm levelling, one of the
processes said to be especially prevalent in situations of language obsolescence (see
section 1).8
However, young speakers are not the only group associated with a preference for
bimorphemic 1SG.OBL and 1SG.GEN forms. The small SLP-speaking community from
Jaffna also shows a clear preference for bimorphemic pronominal forms, but, in this case,
the base selected by the case-suffix is not nominative: it is most often oblique (in 1SG.OBL
paami-pa/paaim-pa and 1SG.GEN paami-su), and, on two occasions only, genitive (in
1SG.GEN minha-su). What is interesting in this case is that, to some extent, it also
constitutes a case of paradigm levelling (in the sense that the bimorphemic nature of most
SLP non-nominative pronouns is extended to the 1SG domain), but introduces an
innovation: the selection of a non-nominative base. This innovation, we argued above, is
not random and reflects the influence of Tamil, a case of adstrate transfer which appears
to impact the Jaffna community more than speakers elsewhere, including the under-40s
mentioned in the previous paragraph. The fact that SLP language loss is particularly
advanced in Jaffna is perhaps implicated in this development, as it combines two of the
types of change said to be reinforced by obsolescence (see section 1): paradigmatic
regularization and permeability to the structural influence of the community’s dominant
language.
This brings us to yet another specificity of the Jaffna community: its impoverished
system of honorificity as encoded in personal pronouns. To be clear, in the absence of
any records prior to our own documentation, it is impossible to determine whether Jaffna
SLP ever had a system of honorificity as robust as what we find in the Eastern varieties.
However, there are reasons to suppose it did: on the one hand, of course, the prevalence
of such a system in the other modern varieties of SLP, but also the hints of honorificity
in the 19th-century descriptions of SLP – which were produced at a time when the
language was more vital in Jaffna and across the island; on the other hand, the fact that
8
The fact that one of these speakers is a 12-year old raises the question of whether language acquisition
could be involved in this type of paradigm regularisation – the rationale being that, in the naturalistic
acquisition of SLP, the introduction of a suppletive form in the paradigm may be preceded by a stage in
which the regular, more transparent bimorphemic form is dominant. However, given that a 12-year-old is
not a young child, and in the absence of longitudinal studies of SLP acquisition, this question cannot be
resolved.
26
the more limited set of 3rd person pronouns used in Jaffna contains forms which, in the
Eastern varieties, are associated with both HON and NHON; and, finally, the fact that
honorificity distinctions are a conspicuous feature of the major languages of Sri Lanka,
Tamil (Schiffman 1999:115-118) and Sinhala (Chandralal 2010:267-272). If this is
correct, then the Jaffna variety effectively lost its previous system of honorificity, which
constitutes an instance of stylistic shrinkage, also associated with situations of language
obsolescence (see section 1). Interestingly, however, the influence from the dominant
language invoked above in connection with the formation of non-nominative 1SG
pronouns is not seen to operate in this case, since the Tamil model has not prevented the
loss of honorificity encoding in SLP.
6. Concluding remarks
Our study of modern SLP personal pronouns has revealed that even such a narrow domain
of grammar can be characterized by plenty of variability. In the case of the data included
in the DSLP corpus, the nature of the observed variation ranges from phonetic alternation
(as in the case of segmental differences between pairs of forms such as osiir and esiir) to
the effects of diachronic processes of variant competition and substitution (as in the case
of the R+ and R- forms of 2PL and 3PL pronouns); from strategies of paradigm
regularization resulting in morphologically different pronominal forms (as in the cases of
the opposition between suppletive and analytic pronouns) to differences in pragmatic
resources and choices (as with the encoding of honorificity).
As expected, not all variants encountered in the corpus are equally prevalent.
However, it is important not to ignore those that have a limited distribution, not only in
the interest of descriptive accuracy, but also because they can be enlightening. In the case
of our study, some of the more circumscribed instances of variation analyzed have turned
out to be associated with groups of speakers characterized by significant language loss,
which, in our interpretation, allows us to observe the impact of obsolescence on the
structure of endangered languages and on the degree of variability of documentary data.
In this study, one such group was defined by the variable of age (the younger
generations of SLP-speakers), while another was defined geographically (the SLPspeakers from the city of Jaffna). The fact that this is the first study of SLP to include data
from Jaffna, and that the Jaffna variety has, in recent times, developed in conditions of
isolation from the wider SLP-speaking community and of advanced language shift,
resulted in new forms and new insights that complexify the overall description of modern
SLP personal pronouns. However, it is important not to lose sight of the fact that much
of the variation discussed here was observed not among the most isolated or least fluent
sections of the speech community, but in the speech of highly fluent speakers. This
27
conveniently demonstrates the extent to which descriptive homogeneity is often a byproduct of data limitations and draws attention to the fact that, if possible, any language
documentation and description endeavor should aim to constitute a sociolinguisticallydiverse and geographically-encompassing pool of consultants, regardless of the actual
size or perceived homogeneity of the speech community.
Abbreviations
1 = first person; 2 = second person; 3 = third person; ABL = ablative; ACC = accusative;
DAT = dative; DEM = demonstrative; DSLP = Documentation of Sri Lanka Portuguese; EXS
= existential F = feminine; FOC = focus; GEN = genitive; HAB = habitual; HON = honorific;
M = masculine; NEG = negative; NHON = non-honorific; OBL = oblique; OBLG = obligative;
PFV = perfective; PL = plural; PRS = present; PST = past; PURP = purposive; QUOT =
quotative; SG = singular; SLP = Sri Lanka Portuguese.
References
Aikhenvald, Alexandra Y. 2020. Language contact and endangered languages. In
Anthony Grant (ed.), The Oxford Handbook of Language Contact, 241-260. Oxford:
Oxford University Press.
Anon. 1826. O Livro de oraçaõ commum e administraçaõ de os sacramentos, e outros
ritos e ceremonias de a Igreja, conforme de o uso de a Igreja de Inglaterra: Juntamente
com o Psalterio, ou Psalmo de David. London: G. Ellerton & J. Henderson.
Anon. 1851. O evangelho conforme de Santo Mattheos. Traduzido ne Indo-Portugueza.
London: R. Clay.
Anon. 1863. O livro de oraçaõs usado ne greyas de Wesleyanos ne Ilha de Ceylon.
Colombo: Officio Wesleyano.
Bakker, Peter. 2006. The Sri Lanka Sprachbund: The newcomers Portuguese and Malay.
In Yaron Matras, April McMahon & Nigel Vincent (eds.), Linguistic Areas: Convergence
in Historical and Typological Perspective, 135-159. New York: Palgrave.
Benveniste, Emile. 1971. Problems in general linguistics. Coral Gables, FL: University
of Miami Press.
Berrenger. 1811. A grammatical arrangement on the method of learning the corrupted
Portuguese as spoken in India. Colombo: Frans de Bruin.
Bhat, D. N. S. 2004. Pronouns. Oxford: Oxford University Press.
28
Callaway, John. 1818. A vocabulary; with useful phrases, and familiar dialogues; in the
English, Portuguese, and Cingalese Languages. Colombo: Wesleyan Mission Press.
Callaway, John. 1820. A vocabulary, in the Ceylon Portuguese, and English languages,
with a series of familiar phrases. Colombo: Wesleyan Mission Press.
Callaway, John. 1823. A Ceylon-Portuguese and English Dictionary. Colombo:
Wesleyan Mission Press.
Campbell, Lyle & Martha C. Muntzel. 1989. The structural consequence of language
death. In Nancy Dorian (ed.), Investigating Obsolescence: Studies in Language
Contraction and Death, 181-196. Cambridge: Cambridge University Press.
Cardoso, Hugo C. 2007. Linguistic traces of colonial structure. In Eric Anchimbe (ed.),
Linguistic identity in postcolonial multilingual spaces, 164-181. Newcastle: Cambridge
Scholars Publishing.
Cardoso, Hugo C. 2014. Factoring sociolinguistic variation into the history of IndoPortuguese. Revista de Crioulos de Base Lexical Portuguesa e Espanhola 5. 87-114.
Cardoso, Hugo C. 2017. Documentation of Sri Lanka Portuguese. London: SOAS,
Endangered Languages Archive. hdl.handle.net/2196/a542c4b1-8c36-4fd5-ae43e777f87f5983. (26 July, 2019.)
Cardoso, Hugo C., Mahesh Radhakrishnan, Patrícia Costa & Rui Pereira. 2019.
Documenting modern Sri Lanka Portuguese In Pinharanda-Nunes, Mário & Hugo
Cardoso (eds.), Documentation and Maintenance of Contact Languages from South Asia
to East Asia (Language Documentation & Conservation Special Publication nr. 19), 133. Honolulu: University of Hawai’i Press.
Cardoso, Hugo C., Tjerk Hagemeijer & Nélia Alexandre. 2015. Crioulos de base lexical
portuguesa. In Maria Iliescu & Eugeen Roegiest (eds.), Manuel des Anthologies, Corpus
et Textes Romans, 670-692. Berlin: Mouton de Gruyter.
Chandralal, Dileep. 2010. Sinhala (London Oriental and African Language Library 15).
Amsterdam/Philadelphia: John Benjamins.
Clements, J. Clancy. 1990. Deletion as an indicator of SVO → SOV shift. Language
Variation and Change 2. 103-133.
Dalgado, Sebastião Rodolfo. 1900. Dialecto Indo-Português de Ceylão. Lisbon:
Imprensa Nacional.
Fox, William Buckley. 1819. A Dictionary in the Ceylon-Portuguese, Singhalese, and
English Languages. Colombo: Wesleyan Mission Press.
Helmbrecht, Johannes. 2003. Politeness Distinctions in Second Person Pronouns. In
Friedrich Lenz (ed.), Deictic Conceptualization of Space, Time and Person, 185-202.
Amsterdam/Philadelphia: John Benjamins.
29
Ishiyama, Osamu. 2019. Diachrony of Personal Pronouns in Japanese. A functional and
cross-linguistic perspective (Current Issues in Linguistic Theory 344). Amsterdam/
Philadelphia: John Benjamins.
Lyons, John. 1977. Semantics, vol. 2. Cambridge: Cambridge University Press.
Mansfield, John & James Stanford. 2017. Documenting sociolinguistic variation in
lesser-studied indigenous communities: Challenges and practical solutions. In Kristine A.
Hildebrandt, Carmen Jany & Wilson Silva (eds.), Documenting Variation in Endangered
Languages (Language Documentation & Conservation Special Publication 13), 116-136.
Honolulu: University of Hawai'i Press.
Meyerhoff, Miriam. 2017. Writing a linguistic symphony: Analyzing variation while
doing documentation. Canadian Journal of Linguistics/Revue Canadienne de
Linguistique 62(4). 525-549.
McGilvray, Dennis. 1982. Dutch Burghers and Portuguese mechanics: Eurasian ethnicity
in Sri Lanka. Comparative Studies in Society and History 24(2). 235-263.
Nagy, Naomi. 2009. The challenges of less commonly studied languages: Writing a
sociogrammar of Faetar. In James Stanford & Dennis Preston (eds.), Variation in
Indigenous Minority Languages, 397-417. Amsterdam/Philadelphia: John Benjamins.
Nagy, Naomi. 2017. Documenting variation in (endangered) heritage languages: How
and why? In Kristine A. Hildebrandt, Carmen Jany & Wilson Silva (eds.), Documenting
Variation in Endangered Languages (Language Documentation & Conservation Special
Publication 13), 33-64. Honolulu: University of Hawai'i Press.
Newstead, Robert. 1827. Um curto catichismo da Biblia, disposado em quarento divisaõs,
todas as repostas per as perguntas sendo ne as palavras de a Santas Escrituras. Transl.
of W.F. Lloyd. London: J.S. Hughes.
Newstead, Robert. 1852. O Novo Testamento de Nossa Senhor e Salvador Jesus Christo.
Transl. of W.F. Lloyd. Colombo: Wesleyan Mission.
Newstead, Robert. 1871. Cantigas per adoraçaõ publico, em lingua portugueza de
Ceylon. Colombo: Wesleyan Mission.
Nordhoff, Sebastian. 2013. The current state of Sri Lanka Portuguese. Journal of Pidgin
and Creole Languages 28(2). 425-434.
Palosaari, Naomi & Lyle Campbell. 2011. Structural aspects of language endangerment.
In Peter Austin & Julia Sallabank (eds.), The Cambridge Handbook of Endangered
Languages, 100-119. Cambridge: Cambridge University Press.
Pereira, Rui. 2019. Sri Lanka Portuguese: A sociolinguistic perspective. Paper presented
at the 19th Annual Conference of the Association of Portuguese and Spanish-lexified
Creoles & Summer Conference of the Society of Pidgin and Creole Linguistics.
Faculdade de Letras da Universidade de Lisboa, June 17th.
Sasse, Hans-Jürgen. 2001. Typological changes in language obsolescence. In Martin
Haspelmath, Ekkehard König, Wulf Oesterreicher & Wolfgang Raible (eds.), Language
30
Typology and Language Universals; An International Handbook, vol. 2, 1668-1677.
Berlin/New York: Walter de Gruyter.
Sippola, Eeva. 2018. Collecting and analysing creole data. In Wendy Ayres-Bennett &
Janice Carruthers (eds.), Manual of Romance Sociolinguistics, 91-113. Berlin: De Gruyter
Mouton.
Siewierska, Anna. 2004. Person. Cambridge: Cambridge University Press.
Siewierska, Anna. 2013. Gender Distinctions in Independent Personal Pronouns. In
Matthew Dryer & Martin Haspelmath (eds.), The World Atlas of Language Structures
Online. Leipzig: Max Planck Institute for Evolutionary Anthropology.
Schiffman, Harold F. 1999. A Reference Grammar of Spoken Tamil. Cambridge:
Cambridge University Press.
Smith, Ian. (coll.). 1973. Indo-Portuguese (Sri Lanka) (IRS01). PARADISEC.
http://catalog.paradisec.org.au/collections/IRS01 (27 July, 2019.)
Smith, Ian. 1977. Sri Lanka Creole Portuguese phonology. Ithaca, NY: Cornell
University PhD dissertation.
Smith, Ian. 1979a. Convergence in South Asia: a creole example. Lingua 48- 193-222.
Smith, Ian. 1979b. Substrata vs. universals in the formation of Sri Lanka Portuguese.
Papers in Pidgin and Creole Linguistics 2. 183-200.
Smith, Ian. 1984. The development of morphosyntax in Sri Lanka Portuguese. In Mark
Sebba & Loreto Todd (eds.), York papers in Linguistics. York: University of York.
Smith, Ian R. 2013. Sri Lanka Portuguese. In Susanne M. Michaelis, Philippe Maurer,
Martin Haspelmath & Magnus Huber (eds.) The Survey of Pidgin and Creole Languages,
Vol. II (Portuguese-based, Spanish-based, and French-based Languages), 111-121.
Oxford: Oxford University Press.
Smith, Ian. 2016. The earliest grammars of Sri Lanka Portuguese. Papia 26(2). 237-281.
Tadmor, Uri. 2009. Loanwords in the world’s languages: Findings and results. In Martin
Haspelmath & Uri Tadmor (eds.), Loanwords in the World’s Languages: A Comparative
Handbook, 55-75. Berlin: De Gruyter Mouton.
Tomás, Maria Isabel. 1992. Os Crioulos Portugueses do Oriente; Uma Bibliografia.
Macau: Instituto Cultural de Macau.
31
Authors’ addresses
Hugo C. Cardoso
Universidade de Lisboa
Faculdade de Letras
Alameda da Universidade
1600-214 Lisboa
Portugal
Patrícia Costa
Universidade de Lisboa
Faculdade de Letras
Alameda da Universidade
1600-214 Lisboa
Portugal
hcardoso@letras.ulisboa.pt
patriciacosta1@campus.ul.pt
32
Hugo Cardoso