Next Article in Journal
On the Licensing Condition on Sluicing: Evidence from Japanese
Previous Article in Journal
Pronoun Mixing in Netherlandic Dutch Revisited: Perception of ‘u’ and ‘jij’ Use by Pre-University Students
Previous Article in Special Issue
Infinitival and Gerund-Participial Catenative Complement Constructions in English World-Wide
 
 
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Article

The Sociolinguistics of Quotatives in Sri Lankan English: Corpus-Based Insights

Department of English, Justus Liebig University Giessen, 35394 Giessen, Germany
Languages 2025, 10(9), 236; https://doi.org/10.3390/languages10090236
Submission received: 20 June 2025 / Revised: 5 September 2025 / Accepted: 11 September 2025 / Published: 18 September 2025

Abstract

This paper examines the quotative system of Sri Lankan English. Quotatives are identified in face-to-face conversations in the Sri Lankan component of the International Corpus of English. The use of kiyala indicating and following quoted material has been transferred from Sinhala, one of the indigenous languages of the country, into Sri Lankan English. Together with the occurrence of complementising that, the localisation of the Sri Lankan English quotative system is evident. Special emphasis is given to the choice between BE like and SAY, the by far most frequent quotative forms in the informal spoken data analysed. They are annotated with established structural (e.g., content of the quote or tense) and sociobiographic variables (e.g., age and gender of the speaker) apparent from earlier quotative research, but also with new ones (e.g., quote length or speakers’ stays abroad or media exposure to particular varieties of English). Via a generalised linear mixed-effects model tree implementing the latest methodological suggestions for classification trees, it is found that BE like is favoured over SAY in Sri Lankan English with younger speakers—particularly when the conversation took place after 2015 and events are narrated using the historical present.

1. Introduction

This study is the first one to systematically explore the Sri Lankan English (SLE) quotative system. It investigates face-to-face conversations as represented in the Sri Lankan component of the International Corpus of English (ICE-SL; Bernaisch et al., 2019). It sets out to identify prominent quotatives in SLE and to understand which structural and sociobiographic variables guide the choice of one quotative over another in a given communicative setting in Sri Lanka. Against this background, the extent to which the SLE quotative system features regionally distinct characteristics is also scrutinised, and features of language contact between the indigenous languages of Sri Lanka and English are profiled. As ICE-SL features extensive sociobiographic information on its speakers, including often-documented sociolinguistic variables like age or gender, but also information on ethnicity, stays abroad, or American and British media use, which is not readily available in most corpora, this paper is in a position to statistically weigh the effects of said variables on quotative choices. Relevant sociolinguistic questions like ‘Does the frequent consumption of American English media trigger a preference for BE like with SLE speakers?’ or ‘Are quotative choices by SLE speakers influenced by whether they had longer stays abroad’ can be discussed on the basis of statistically reliable evidence (instead of intuition-based impressions). This statistical evidence is provided in the form of a (here tree-based) multifactorial model as is currently standard in corpus-based research, although earlier research on quotatives may have relied on less complex—and occasionally less trustworthy—statistical designs. In Section 2, theoretical observations on quotatives in the World Englishes paradigm are offered and SLE is presented from historical, structural, and sociolinguistic angles. In Section 3, the corpus data and the methods of data extraction, data annotation, and statistical modelling are presented. Section 4 outlines variety-specific features of the SLE quotative system and focuses on how SLE speakers make the choice between the most dominant quotatives BE like and SAY in a given communicative context. Section 5 discusses the findings of the paper at hand in the light of earlier research and offers some concluding remarks.

2. Theoretical Background

Quotatives such as SAY, THINK or BE like introduce “the recreation of speech, thought, action, sound, or gesture” (D’Arcy, 2012, p. 347) to a spoken or written text. For instance, in She said ‘I don’t think that’s a good idea’, SAY is the quotative that introduces material quoted from another speaker. Particular academic interest has been shown to different formal realisation of quotatives, but also to factors that may influence speakers to choose one quotative over another in a given communicative setting. Research into quotative forms captures then novel forms such as this is + [speaker] as evident in London (Cheshire et al., 2011; Fox, 2012), the wave-like short-term diachronic frequency profile of BE all in California (Buchstaller et al., 2010; Rickford et al., 2007) or the emergence of GO in various parts of the English-speaking world (Buchstaller, 2006; Butters, 1980). Here, D’Arcy (2015) provides an insightful discussion of language-internal and -external factors to be considered in such analyses of quotatives. To understand the contextual constraints that influence formal quotative choices, the literature on BE like is instructive. It documents that (a) the nature of the quoted content (Buchstaller & D’Arcy, 2009; D’Arcy, 2010; Tagliamonte & Hudson, 1999), (b) grammatical factors like the person of the subject (e.g., Barbieri, 2005; Tagliamonte & Hudson, 1999) or tense (e.g., Blyth et al., 1990, p. 218; Singler, 2001, p. 271), and (c) the sociobiographic factors age (e.g., Barbieri, 2009; Blyth et al., 1990) and gender (e.g., Buchstaller, 2008; Ferrara & Bell, 1995) affect BE like frequencies. Against this background, D’Arcy’s (2012, p. 345) appreciation of BE like as “a robust heuristic in the effort to develop an empirical theory of language change, providing key insights into such issues as age-grading, lifespan change, and the incrementation of change, as well as allowing detailed exploration of linguistic, social, and geographic facets in the diffusion of change” is certainly in place. At the same time, the “geographic facets” (D’Arcy, 2012, p. 345) explored in studies on quotative systems are often confined to native-speaker contexts while studies dedicated to quotative choices in second-language varieties of English are still comparatively limited. As the present paper is also situated in the World Englishes paradigm, Section 2.1 outlines central findings concerning quotatives in the Kachruvian Outer Circle. Section 2.2 is a selective synopsis of research into SLE, the South Asian variety of English examined in the present paper.

2.1. Quotatives in World Englishes

The first study to take the investigation of quotatives into the sphere of second-language varieties of English is Höhn (2012), where Jamaican and Irish English are compared via the respective components of the International Corpus of English (ICE; Greenbaum, 1996). Her comparison targets the quotatives BE like, GO, and SAY, which were extracted from the datasets through concordances of the respective word forms accompanied by careful reading of the corpus texts. While the innovative quotatives BE like and GO are employed in Irish English, Jamaican English speakers use BE like, but not GO, as evident from the by now relatively dated Jamaican ICE component. Each corpus example has been annotated with regard to the register and year in which it occurred, the grammatical person of the subject in the clause where the quotative was found, the content of the quote in terms of whether it represented direct speech or internal dialogue, and the gender of the speaker. A multifactorial model of quotative choices indicates that “Jamaican women are more likely to use be like and […] this is a strong conditioning effect. Furthermore, the innovative quotative is favoured for internal dialogue in this dataset” (Höhn, 2012, p. 286).
Based on her multivariate analysis of BE like in Jamaican English, Bogetić (2014) is able to add that BE like tends to co-occur with historical present, i.e., the use of present tense in an otherwise past-tense narrative to draw a listener’s or reader’s attention, and confirms that Jamaican speakers of English rather employ BE like to report thought than speech. In contrast, in Irish English, BE like tends to be preferred by male speakers in clauses where the subject is in the first person (Höhn, 2012, p. 285).
D’Arcy (2013) investigates quotatives also on the basis of data drawn from ICE with a focus on Hong Kong, Indian, Kenyan, Philippine, and Singapore English. For these varieties, she (2013, p. 497) attests that quoted speech in contrast to imagined content constitutes the vast majority of quoted content. Concerning individual quotative forms, it is concluded that “say is consistently the most frequent quotative overall […], while think is in all cases a minority form, consistently (among) the least frequent of all the verbs used. Thus, the elevated rates of inner state reporting that are found in the contemporaneous Inner Circle varieties simply do not manifest in the Outer Circle ICE materials” (D’Arcy, 2013, p. 497). The set of quotative forms in second-language varieties is also argued to be more restricted than that of native varieties, since forms other than ASK, BE like, GO, SAY, TELL, THINK and zero quotatives cannot be found in the Outer-Circle ICE components studied. BE like is profiled as a low-frequency quotative in Outer-Circle Englishes in general and as absent from the Indian data (D’Arcy, 2013, p. 497). In contrast, ASK and TELL are dominant quotatives across the World Englishes examined and the complementiser that following a quotative to introduce direct speech is used notably more often in second- than in first-language contexts (D’Arcy, 2013, p. 498). D’Arcy (2013, p. 498) summarises that “[t]he nearly categorical tendency to report speech rather than other types of content, the restricted repertoire, and the ability to collocate ‘that’ with verbs of direct quotation suggest that the ecology of quotation functions distinctly in Outer Circle Englishes than it does in Inner Circle varieties” (D’Arcy, 2013, p. 498). Still, Hansen (2021) reminds us that the Outer-Circle insights D’Arcy (2013) presents rely on texts in ICE that date back to the 1990s. It follows that potential more recent changes and advances in the use of the innovative quotatives in particular might have remained unexplored in this study.
In this light, Davydova’s (2016) study on quotatives in Indian English is a most welcome contribution in that it incorporates data from ICE India collected from 1990 to 1996, but also more recent texts from the Hamburg Corpus of Non-Native Varieties of English, which features authentic Indian English material from 2007 to 2011. This short-term diachronic setup allows Davydova (2016) to observe that SAY, the most dominant quotative in Indian English in the 1990s occurring in almost 2 out of 3 quotative cases, is still the most frequently used quotative in the data from the late 2000s. At the same time, its relative frequency of occurrence has gone down to being used in approximately 1 out of 3 quotative cases. This is related to noteworthy short-term diachronic increases in zero quotatives and BE like as well as okay fine, the latter of which is illustrated in (1) and (2).
(1)
they might would probably ask me that okay (,) fine, ‘I need five thousand rupees’. (HCNVE-India IE35; Davydova, 2016, p. 185)
(2)
So that used to be okay (,) fine, ‘We are sitting in an English class’. (HCNVE-India IE35; Davydova, 2016, p. 185)
In (1), it might still be possible to view ASK in combination with me and the complementiser that as the quotative structure, relegating okay fine to the status of the onset of the quoted material. Still, in (2), it is hard to argue against the quotative role of okay fine since no alternative quotative forms are realised in this speaker turn. For Davydova (2016, p. 185), these examples illustrate the grammaticalization of quotatives, since she hypothesises that quotative markers “start out as features signalling the onset of the quoted material and gradually become more and more entrenched within the quotative template until completely conventionalised”—a hypothesis which warrants further investigation with novel quotative forms also in World Englishes.
Her multifactorial analysis of the contexts that may attract quotative SAY rests on the structural predictors person of grammatical subject, tense, and quoted material, and the sociobiographic predictors age, gender, and occupation. She summarises that person of the grammatical subject and tense were important constraints for quotative SAY in the Indian English of the 1990s, while the type of quoted material and tense were relevant in the late 2000s.
In a related study on quotatives in Indian English in the late 2000s, Davydova (2019) profiles the structural predictors grammatical subject, tense, mimesis for whether quoters imitate the original speaker’s voices and quoted material, as well as the sociobiographic factors gender, dominant language context, and mass media exposure as statistically significant for speakers’ inclination to employ BE like as a marker for a quote. Specifically in terms of mass media exposure, Davydova (2019, p. 128) reports that “speakers watching English-speaking films and TV series on a regular basis are also the most likely users of be like”. With a view to the content of quote, D’Arcy (2013) states for her World Englishes data from the 1990s that “the rates of quoted speech are extremely high overall. In this respect, the ICE data resemble the quotative system of late nineteenth- and early twentieth-century Inner Circle English […]. It is highly circumscribed to the reporting of speech”. In contrast, the Indian English data from the late 2000s indicate a development towards BE like preferring imagined/mental quoted material over spoken material (Davydova, 2019, p. 130).
Complementing Indian English sociolinguistic interview data from 2007 to 2014 with similar data from German speakers of English collected from 2014 to 2016, Davydova (2021, p. 182) establishes through a mixed-effects analysis that “mimesis, quote type, grammatical subject, and tense are genuine predictors of be like occurrence”. BE like is preferred with first-person subjects in historical-present tense when the quoted material is thought and the sound characteristics of the original material are imitated in the quote (Davydova, 2021, p. 182). Notably, though, the difference in regional background, i.e., India as a prototypical second-language (ESL) territory versus Germany with foreign-language (EFL) status for English, turns out to be non-significant, which Davydova (2021, p. 183) profiles as an indication that “ESL/EFL speakers have adopted the variable grammar governing the use of be like in native English with a remarkable degree of precision.”
Hansen (2021) is the first study that puts exclusive focus on the quotative system in an African variety of English, namely Ghanaian English. In the Ghanaian ICE component, SAY constitutes 49% of quotatives followed by BE like (17%) and zero quotatives (15%). The combination of a quotative with the complementiser that is also evident from the Ghanaian data, which lends support to D’Arcy’s (2013) perspective that complementising that—in Ghanaian English in combination with, e.g., TELL and REALISE—is a feature of second-language varieties of English. Localised Ghanaian quotative forms featuring variants of the complementiser se prominent in Ghanian Pidgin English to introduce direct speech ‘after verbs of cognition, perception, or saying’ (Huber, 1999, p. 188) are exemplified in (3) and (4).
(3)
told you seh Nancy you have a thing for this guy one <ICE-GHA:S1A-040> (Hansen, 2021, p. 38)
(4)
the guinea pig ask me say ah but where is the surgical knife <ICE-GHA:S1A-015> (Hansen, 2021, p. 38)
Hansen (2021) studies BE like in Ghanaian English through the structural factors quoted content, person of the grammatical subject, and tense, in addition to the sociobiographic variables age and gender. She (2021, p. 39) finds that “the effect of the linguistic factors is in the expected direction, with a preference for be like with inner monologues and the historical present”. In complementation to Davydova (2019), this is further evidence for quotative BE like no longer being strongly associated with quoting speech in second-language varieties. In terms of sociobiographic factors, BE like becomes less likely in Ghanaian English as speaker age increases, but speaker gender did not turn out to significantly influence BE like choices, as might have been expected from earlier studies into Inner-Circle Englishes.
Conversely, Deuber et al. (2021) detect a significant effect of age alongside that of gender in Trinidadian English, in that BE like is preferred by younger female speakers. Still, their multifactorial model does not consider any other structural or contextual predictors that earlier research profiled as significant, which might unduly boost the importance of age and gender and related interpretations.
Earlier multifactorial findings are summarised in Table 1. Ticks indicate that a given variable has been analysed in a particular study and adjacent pluses show that this variable has been profiled as significant in the study concerned.
Concerning the structural predictors, the “classic factors of grammatical person and content of the quote” (Tagliamonte & D’Arcy, 2007, p. 203) are featured in all the multifactorial studies on quotatives in World Englishes, with the exception of Deuber et al. (2021). Tense is also featured as a predictor in most studies, but mimesis, i.e., the imitation of the sound characteristics of the original speaker of the quoted material, has so far only been integrated into multifactorial statistical models for quotatives in Davydova (2019, 2021)—probably due to the fact that mimesis must rely on phonetic/phonological information rarely available in larger corpora. In terms of sociobiographic factors, gender has been included systematically, while age has also been repeatedly featured. A temporal dimension has been offered in Höhn (2012) and Davydova (2016), but education, occupation, degree of media exposure, and dominant language context have so far not been integrated repeatedly into the study of quotatives in the World Englishes paradigm—which is also because these types of information only seldom come with corpus data that is available for second-language varieties. In order to allow this study to connect with earlier findings and to give adequate statistical weight to predictors known to be potentially important for quotatives, the present study adopts all predictors from Table 1 to the exception of region because the focus is solely on Sri Lanka. Further, it explores additional predictors to understand whether they can significantly add to explanations of variability in quotative choices. All predictors of the present study and their operationalisations are documented in Section 3.2.
The quotative systems of a number of World Englishes have consequently already been explored. This appears particularly relevant since “transformation under transfer” (Meyerhoff, 2003, p. 342) might be in effect when quotatives are employed in non-Inner-Circle contexts: “the specific details of its [=BE like’s] social and functional constraints are re-created by localised groups of speakers, who adopt and routinize the newcomers in a locally specific way” (Buchstaller & D’Arcy, 2009, p. 323).
A variety whose quotative usage as well as related means of localisation have so far remained undocumented is SLE, the variety in focus in the present study.

2.2. Sri Lankan English

This section provides a brief history of SLE and corpus resources available for the study of this South Asian English.1 The hitherto relatively limited insights into its quotative system will also be presented.
The first contact between English and then Ceylon dates back to 1796 when the British East India Company sought to establish business relations with the locals. In 1802, the island became part of the British Empire (de Silva, 1981, pp. 210–211). Initially, the colonial rulers did not show much interest in the education of the local people, but realised that their administration would eventually benefit from effective means of communication with them. With this goal in mind, the Colebrooke–Cameron Commission of 1831/1832, seeking to educate a small elite in the English language and British culture, represents the beginning of the attempt to Westernise Sri Lankan society—primarily through the teaching of the English language (Yogasundram, 2008, p. 238). The implementation of the respective recommendations by the commission marks the beginning of the development of SLE into a variety of English in its own right, since a small group of locals was then systematically introduced to the English language, instantiating a steady contact between English and the major local languages Sinhala, Tamil, and Malay.
As a result of this expansive language contact between Sri Lanka’s indigenous languages and English, but potentially also because of conceivable epicentral influences from its large neighbour variety Indian English (Gries & Bernaisch, 2016) and other creative forces driving linguistic innovation (Bernaisch, 2015), SLE and its linguistic structures have been nativised across all structural levels of language organisation. At the sound level, (combinations of) consonants and vowels are produced in specifically local ways (Gunesekera, 2005; Senaratne, 2009), SLE lexis has borrowed extensively from Sinhala and Tamil (e.g., AMMA for mother), but also features vocabulary items that may be deemed archaic (e.g., DAMSEL for young woman) or formal (e.g., DETRAIN for getting out of a train) from a contemporary British English perspective (Meyler, 2007). Lists of grammatical innovations in SLE like valency reduction as in I like or Can you believe? or the prominence of topicalisation as in A tremendous speech she gave are available in Gunesekera (2005) or Meyler (2007). Also, the semantics of certain structures have been localised in that, e.g., the particle verb PASS out can be used to denote graduation from a school (Meyler, 2007).
The formal status of English as defined in the Sri Lankan Constitution is that of a “link language” (Coperahewa, 2009, p. 93), which is probably meant to serve as a comparatively neutral medium for communication between the local ethnicities inhabiting the island. In terms of sociolinguistics domains of use, “English is a dominant language of local print and electronic media such as the radio, television, newspapers, and magazines” (Ekanayaka, 2020, p. 340), but is, of course, also prominent online. Despite the re-integration of an English-medium stream in public schools offering—in theory—English-medium education to every Sri Lankan student, the reality evident from the 2015 survey by the Sri Lankan Ministry of Education is that only 7.3% of government-school students received any form of instruction in English. While the reasons for this may very practically be rooted in the lack of teachers competent enough to teach (in) English, the result is that the English language has been and continues to be a marker of social prestige and status for a small Sri Lankan elite (Ekanayaka, 2020, pp. 340–341). These historical, structural, and sociolinguistic perspectives on SLE indicate that this postcolonial English can generally be regarded as an endonormatively stabilised variety in the framework of Schneider (2007), although some of the respective parameters such as the stabilisation of a new variety or the relative homogeneity of local norms will have to be empirically scrutinised when adequate databases become available for study (Bernaisch, in press).
The empirical study of SLE has been facilitated through a growing number of corpora. The Sri Lankan component of the South Asian Varieties of English (SAVE; Bernaisch et al., 2011) Corpus was the first larger corpus resource of SLE. It features approximately 3,000,000 words of newspaper writing produced at the two largest national English-medium newspapers, the Daily Mirror and the Daily News, with texts from 2001 to 2007. The Sri Lankan component of the 2020 Update of the South Asian Varieties of English (SAVE2020; Bernaisch et al., 2021) Corpus is—as the name suggests—a re-iteration of the original SAVE-SL component in the sense that it adopts the compilation principles of SAVE to collect texts from the Daily Mirror and the Daily News. The text featured in the Sri Lankan component of SAVE2020 were published in 2019 and 2020, which enables short-term diachronic perspectives.
The Sri Lanka English Newspaper Corpus (SLENC; DHLab—Department of English, 2022) follows the corpus design of SAVE and SAVE2020—also in that the Daily Mirror and the Daily News are used as its sources—and makes 31.8 million words of SLE newspaper writing from the period 2015–2018 available. The News on the Web (NOW; Davies, 2016) Corpus also offers Sri Lankan English-medium newspaper articles from 2010 to the present, which currently amount to roughly 47 million words. The Corpus of Global Wed-based English (GloWbE; Davies, 2013) also features a Sri Lankan component comprising circa 47 million words from Sri Lankan websites and blogs.2
The web-based resources discussed are restricted to SLE writing. The Sri Lankan component of ICE (ICE-SL) is so far the only dataset featuring authentic spoken and written texts across genres associated with different degrees of formality. Together with a comparatively wide range of sociobiographic speaker information, this resource facilitates the exploration of linguistic innovations in SLE and their potential sociolinguistic driving forces. As this study is rooted in ICE-SL, more details are presented in the methodological Section 3.1.
As regards quotatives in SLE, only a handful of comments can be found. In his dictionary entry for kiyala, Meyler (2007, p. 139) comments that it “is a Sinhala word which marks the end of a quotation, reported statement, indirect question, etc., and which is sometimes used in colloquial SLE in the same way”. (5) and (6) are taken from the Dictionary of Sri Lankan English (Meyler, 2007, p. 139) to illustrate the use of kiyala.
(5)
He called to tell he might be getting late kiyala.
(6)
Customer is going abroad kiyala.
The function of kiyala to signal the reporting of what another speaker said is particularly evident from (5). Senaratne (2009, pp. 204–205) adds that kiyala tends to be placed at the end of clauses. Still, what should not be overlooked is that (6) does not show the use of kiyala in the context of a quotative indicating the re-creation of what another speaker produced, which is why it will have to be established to what extent kiyala is indeed associated with quotative use. The only other statement regarding quotatives in SLE relates to the frequency of BE like, which is supposed to be used extremely rarely in this South Asian English (Kortmann et al., 2020).
In this light, the present paper sets out to empirically explore the quotative system in SLE. The central research questions to be answered are as follows:
  • Which forms are used as quotatives in SLE and how frequent are they?
  • Which structural and sociobiographic factors significantly influence quotative choices in SLE and what effects do these factors exert?

3. Materials and Methods

This methodological part presents the corpus data in Section 3.1 and describes how quotatives were extracted and annotated in Section 3.2. Section 3.3 discusses the choice of the statistical modelling technique for the data at hand.

3.1. Corpus Data

As mentioned in Section 2.2, ICE-SL forms the empirical basis of this study. The texts sampled in ICE-SL mirror “‘educated’ or ‘standard’ English. […] The people whose language is represented in the corpora are adults (18 or over) who have received formal education through the medium of English to the completion of secondary school” (Greenbaum, 1996, p. 6). More precisely, the speakers featured in an ICE component “were born in the country concerned, or if not, […] moved there at an early age and received their school education through the medium of English in that country” (Nelson, 1996, p. 28). Consequently, an ICE component seeks to document that type of local variety of English within a given English-speaking country that is most likely to be established as the national norm as soon as means of standardisation such as codification, political promotion, etc., are set in motion.
For this study, quotatives were analysed in the 90 face-to-face conversations in ICE-SL, since speaker information salient for the study of quotatives is most systematically available for this subset of texts in ICE-SL. Given that ICE-SL is a sample text corpus with a text length of approximately 2000 words each, the total number of words analysed amounts to roughly 180,000 words. The addition of other spoken and written material featured in ICE-SL would have been desirable to establish, e.g., the degree to which the usage patterns of certain quotatives may vary in the light of different degrees of formality within the spoken mode and what kind of differences may exist in quotative use across speech and writing. Although the identification of quotatives in these ICE-SL texts not considered in this study would have certainly been feasible, the often-unavailable metadata for speakers/writers in terms of their, e.g., age or gender, i.e., in terms of factors that earlier studies profiled as central (see Table 1), would have hindered the statistically reliable evaluation of the driving factors guiding quotative choices. As the other SLE corpus resources introduced in Section 2.2 do not consistently offer any type of speaker information either, the comparatively informal face-to-face conversations (Xiao, 2009)—also as a potential cradle of linguistic innovation—appear a reasonable starting point for the description of the SLE quotative system.

3.2. Data Extraction and Annotation

The quotative studies in World Englishes discussed in Section 2.1 successfully identified a large number of different quotative forms. While it would be expected that some of these quotative forms—particularly high-frequency forms shared across different varieties—are prominent in SLE as well, it could be that SLE has developed variety-specific quotatives not (yet) employed by speakers outside Sri Lanka. Further, as zero quotatives have repeatedly been documented in the quotative systems of World Englishes (e.g., Davydova, 2016, 2019; Hansen, 2021), creating concordances based on the word forms of quotatives documented in earlier studies would have (a) potentially missed out on novel quotative forms and uses developed in SLE and (b) failed to capture zero quotatives. Consequently, the 90 face-to-face conversations in ICE-SL were read to identify instances where speakers re-create and incorporate into their speech discourse material, i.e., sound or thought in the widest sense, produced verbatim beforehand. In line with Davydova (2016) and Hansen (2021), certain cues like the occurrence of interjections, discourse markers or pauses, but also overt changes in the deictic centre or corpus mark-up indicating the imitation of voices were monitored for quotative identification. In this light, indirect quotations evident, e.g., from shifted deictic expressions (e.g., He said that he would come <ICE-SL:S1A-070#207:1:B>) were not extracted and ambiguous cases where a reading as an indirect quotation was possible (e.g., she said that uh that they send it to a lot of people <ICE-SL:S1A-008#61:1:A>) were also discarded.
622 quotatives were identified with the help of this manual extraction procedure and the respective forms and their frequencies of occurrence are presented in Section 4.1 to document how quotation is contemporarily marked in SLE. Still, BE like and SAY together constitute 500 out of the 622 quotatives (80.39%), which is why they and the choice between them deserves particular attention.
In Section 4.2, the communicative contexts favouring BE like or SAY are profiled.3 It is evident from earlier research that the structural and sociobiographic factors in Table 1 can exhibit statistically significant influences on quotative choices, making their integration into any future statistical investigation mandatory—particularly so because said studies explicitly study BE like and SAY in World Englishes. From the 500 cases of BE like and SAY, 411 were retained because they featured all the sociobiographic speaker information needed/wanted for analysis. The structural and sociobiographic variables to model the choice between BE like and SAY, the ways in which they have been operationalised, and how the quotatives BE like and SAY are distributed across the levels of each variable are outlined here.
  • person_subject encodes the person of the grammatical subject of the quotative. In absolute frequencies, there were 151 first-person, 12 second-person, and 248 third-person subjects. As the number of 12 second-person subjects appeared too low to constitute a variable level in its own right, first- and second-person subjects are treated collectively in the variable level non-third, which contrasts with the variable level third. Third-person subjects have a tendency to be realised with BE like (58.87%; 146 out of 248 examples), but this tendency is even stronger for non-third-person subjects (71.17%; 116 out of 163 examples).
  • content_quote captures the nature of the quoted material. Oral stands for quoted speech (e.g., she said I smiled because I was embarrassed <ICE-SL:S1A-003#108:1:B>), mental for the re-creation of imagined content as well as feelings, i.e., generally material that was previously not explicitly uttered (e.g., I felt like okay if I was meeting you guys I would have kept a bit of time <ICE-SL:S1A-084#90:1:B>), and indigenous for passages—explicitly uttered or not—using predominantly Sinhala or Tamil instead of English (e.g., he’s asking uh ewunata aunty hadanawada coffee <ICE-SL:S1A-003#143:1:B>). Indigenous material is always accompanied by BE like (100%; 12 out of 12 examples) while mental (84.85%; 28 out of 33 examples) and oral material (60.66%; 222 out of 366 examples) also shows a preference for BE like.
  • tense encodes the tense of the verb phrase in which the quotative occurs. The levels of tense are past, historical present, and non-past, which includes present and future tense, the latter of which was only attested in two cases and was thus combined with the present-tense cases into one level. Modal verbs occurred extremely rarely in the data and have been assigned to tense categories based on their morphological realisation and context, but do not—because of their low frequencies of occurrence—constitute a category in their own right. Corpus examples were considered instances of historical present when the quotative occurred in present tense in an otherwise past-tense narrative. In case the tense is historical present, BE like (95.38%; 62 out of 65 examples) is the dominant choice. In non-past (60.87%; 42 out of 69 examples) and past (57.04%; 158 out of 277 examples) contexts, BE like is still dominant, but not as prevalent as in the historical present.
  • quote_length documents the logged length of the quoted material in number of words. Quotations introduced with BE like are on average slightly shorter (mean = 1.67, sd = 0.86) than those introduced by SAY (mean = 1.7, sd = 0.94).
  • number_subject documents whether the grammatical subject of the quotative is singular or plural. Both singular (62.91%; 229 out of 364 examples) and plural (70.21%; 33 out of 47 examples) subjects show a tendency towards BE like.
  • age represents the logged speaker age when the quotative was produced. On average, speakers are younger when they use BE like (mean = 3.13 (22.87 non-log-transformed age), sd = 0.12) than when they use SAY (mean = 3.36 (28.79 non-log-transformed age), sd = 0.45). The non-log-transformed age values range from 18 to 73.
  • education is a variable with three levels: GCE, BA, and BA+. GCE stands for speakers whose highest educational qualification is the General Certificate of Education (GCE). The GCE Ordinary Level is usually awarded in Sri Lanka over the course of grades 10 and 11 of senior secondary school and the GCE Advanced Level during grades 12 and 13 at collegiate school level. The variable level GCE thus captures speakers with their highest academic achievement at the advanced school level. BA stands for speakers with a BA university degree and BA+ for speakers with more advanced university degrees such as a Master’s or a PhD. Across these educational levels, BE like (GCE = 61.38% (116 out of 189 examples), BA = 65.91% (116 out of 176 examples), BA+ = 65.22% (30 out of 46 examples)) is consistently more frequent than SAY.
  • ethnicity captures the ethnic group corpus speakers identify with. The majority of corpus speakers are Sinhalese (68.17%), 15.33% are Tamil, and the remaining corpus speakers belong to other ethnic groups (16.55%). The Sinhalese (67.86%; 190 out of 280 examples) and speakers of other (67.65%; 46 out of 68 examples) ethnic belonging tend towards BE like, whereas Tamil speakers prefer SAY (58.73%; 37 out of 63 examples).
  • gender encodes the gender group the informants associate themselves with. Female corpus speakers’ preferred quotative choice is BE like (68.65%; 219 out of 319 examples), while male corpus speakers display a slight tendency towards SAY (53.26%; 49 out of 92 examples).
  • home_language describes the language that corpus speakers profiled as the language predominantly used in their homes. Speakers with English (62.44%; 128 out of 205 examples) or Sinhala (73.33%; 132 out of 180 examples) as the main language at home employ BE like more often than SAY, while SAY (92.31%; 24 out of 26 examples) is particularly dominant when Tamil is the home language.
  • ame_media_frequency describes whether corpus speakers consume American English media on a daily or non-daily basis. While BE like is preferred under both conditions, daily consumers of American English have a stronger inclination towards BE like (67.07%; 167 out of 249 examples) than non-daily consumers (58.64%; 95 out of 162 examples).
  • bre_media_frequency captures whether corpus speakers expose themselves to British English media on a daily or non-daily basis. BE like is the dominant variant in both cases, with daily consumers of British English displaying slightly lower frequencies of BE like (62.28%; 175 out of 281 examples) than non-daily consumers (66.92%; 87 out of 130 examples).
  • occupation classifies the corpus speakers into three groups, that is people with a non-teaching profession in contrast to teachers and students. The preference for BE like is strongest with students (70.79%; 206 out of 291 examples), but also notable with teachers (68.75%; 22 out of 32 examples). In contrast, speakers with a non-teaching profession prefer SAY (61.36%; 54 out of 88 examples) over BE like.
  • year_of_recording documents the year in which the quotative was produced to potentially observe a change in quotative preference in the course of the compilation period of ICE-SL. On average, BE like is employed later (mean = 2014.73, sd = 1.68) than SAY (mean = 2014.52, sd = 1.63).
  • stays_abroad describes whether a corpus speaker spent more than six months at a time outside Sri Lanka. Corpus speakers without a stay abroad prefer BE like (69.02%; 205 out of 297 examples) while speakers with a stay abroad use BE like (50%; 57 out of 114 examples) and SAY (50%; 57 out of 114 examples) equally often.

3.3. Statistical Modelling

To evaluate how individual annotated variables influence the choice between BE like and SAY while controlling for the others, a multifactorial statistical model is called for. Of the 254 speakers in the 90 face-to-face conversations in ICE-SL, 106 speakers employed quotatives. Yet, 77 used more than one quotative, with the highest number of quotative use by individual speakers being 17. Consequently, a statistical model for this type of data needs to be able to identify and separately treat potential speaker-related variation in the form of a random effect. Generalised linear mixed-effects model trees (glmertrees; Fokkema et al., 2018) are classification trees that feature the possibility of including such random effects and are thus superior to other classification-tree approaches like conditional inference trees (Hothorn & Zeileis, 2015), treating each observation as independent from one another. In principle, glmertrees are similar to other tree-based approaches in that they arrange observations in subsets of the full dataset in order to make these subsets continuously more homogenous regarding an output variable. Still, in addition to that, glmertrees can partial out random effects in datasets that may manifest themselves in repeated measures, i.e., observations/examples that are not independent from one another because these observations/examples, e.g., were produced by the same speaker, featured the same trigger lemma for a certain construction, etc. Given that repeated measures of these types are a rule rather than an exception in non-experimental linguistic analyses, glmertrees have already been used in corpus-linguistic studies in World Englishes on, e.g., the progressive (Rautionaho & Hundt, 2021), passive voice (Hundt et al., 2024), or parentheticals (Degenhardt, 2025). For a practical guide on how to implement glmertree models in the scripting language R (R Core Team, 2024), please refer to Bernaisch (2022) using pragmatic data on pausing from different World Englishes for illustration. Still, classification trees on occasion do not portray interactions between predictors as significant, which is why interaction predictors were created and integrated into the model to force it to explicitly evaluate and show interaction effects in case they are significant (Gries, 2020). Random forests could have also been used to statistically analyse the quotatives at hand, while mixed-effects models might have posed problems given the large number of potential predictors: “[P]erformance of LMM trees [=linear mixed-effects model trees] is not affected by the number of covariates, whereas the predictive accuracy of LMMs [=linear mixed-effects models] deteriorates when the number of covariates increases, especially when the true interactions are not purely continuous. This indicates that LMM trees are especially useful for exploratory purposes, where there are many potential moderator variables.” (Fokkema et al., 2018, pp. 2028–2029) Given that the current dataset (a) features repeated measures in the form of several speakers contributing various examples and (b) is annotated for a number of predictors whose effects on quotatives have so far not been explored, a glmertree is a viable statistical model for the data at hand.

4. Results

4.1. Quotatives in Sri Lankan English

The quotative system of SLE features a range of different options to introduce quoted material. Table 2 shows the absolute frequencies and proportions of quotative forms in SLE that occurred more than once in ICE-SL sorted in descending frequency. The respective quotative forms are exemplified in (7) to (13). The category ‘Other’ represents the quotatives BE all, EXPLAIN, FEEL, FEEL LIKE, GET all (as in (14)), LABEL, LIKE, MESSAGE, SCREAM, SEND, WONDER, and WRITE that were only attested once in the data, but do not appear to be unique to SLE.
(7)
<ICE-SL:S1A-060#284:1:A> So tenth season they were like if you want us to do it you have to pay us
(8)
<ICE-SL:S1A-001#191:1:A> I said that is the trend then you know this is not a new one
(9)
<ICE-SL:S1A-071#417:1:B> One day she just told me you know you don’t ask for permission anymore
(10)
<ICE-SL:S1A-053#6:1:C> He he asked me can I get a photocopy of this to keep
(11)
<ICE-SL:S1A-007#219:1:B> And then yeah I thought she’ll come at around six-thirty
(12)
<ICE-SL:S1A-034#226:1:B> Or he’ll say okay I’ll drop you home
<ICE-SL:S1A-034#227:1:B> He lives close by
<$A><ICE-SL:S1A-034#228:1:A> Ahh in Dehiwala
<$B><ICE-SL:S1A-034#229:1:B> [zero] So I’ll pick you in the morning on the way to work you can tell me about it
<$A><ICE-SL:S1A-034#230:1:A> but that’s good so you get a ride to work
(13)
<ICE-SL:S1A-025#164:1:B> On his one knee <,> goes down on one knee and proposes to her like <,> and she goes oh my god oh my god you know the usual
Constituting more than every second quotative form, BE like (52.25%) is the dominant quotative in informal spoken SLE. Only SAY with a relative frequency of 28.14% appears to be able to compete with BE like, while the remaining quotatives like TELL, ASK, THINK, zero, and GO, as well as others, are notably less frequent and occupy small niches in the SLE quotative system.
In Davydova’s (2019, p. 125) study on quotatives in Indian English, mimesis, i.e., imitating the original speaker when the quoted material is re-created, is profiled as a relevant factor slightly boosting occurrences of BE like with younger speakers. Examples (14) and (15) show that mimetic quotes are also evident from ICE-SL—here introduced by GET all and BE like.
(14)
<$B><ICE-SL:S1A-003#280:1:B> Not sad she gets all <O>imitation</O>
<ICE-SL:S1A-003#281:1:B> Child you haven’t eaten
(15)
<$A><ICE-SL:S1A-015#242:1:A> then she was like I don’t know <O>imitation</O>
<ICE-SL:S1A-015#243:1:A> And now they’re all on texting terms and everything
Still, mimesis plays a comparatively minor role with SLE quotatives in that it is only featured in 4 out of 622 (0.64%) cases, i.e., three times with BE like and once with GET all. The producers of mimetic quotes are—in the data at hand—female students not older than 23 years with Sinhalese backgrounds. Given these low frequencies, mimesis was not included as a predictor in the multifactorial analysis in Section 4.2.
A similarly marginal characteristic of SLE quotatives is the introduction of quoted material through complementising that. Examples (16) and (17) illustrate its use in SLE.
(16)
<ICE-SL:S1A-010#266:1:B> Ama didn’t say I mean Ama said that you know <,,>
<ICE-SL:S1A-010#267:1:B> I don’t think it’s going to go the way I was thinking it’ll go
(17)
<ICE-SL:S1A-070#87:1:A> I asked him that <,,> tch <,,> where are you
In ICE-SL, complementising that is employed by female and male SLE speakers in their early twenties. What is noteworthy and might warrant further analysis is the question whether complementising that might have been introduced into SLE through language contact between English and Sinhala. All users of that as a complementiser are ethnically Sinhalese and report—to the exception of one speaker who uses English at home—Sinhala as their home language. None of the examples are in any way associated with Tamil ethnicities or Tamil as a language, which, however, might of course also be rooted in data sparsity. In her analysis of first-generation ICE components, D’Arcy (2013, p. 498) contrasts Inner- and Outer-Circle quotative systems based on the tendency of Outer-Circle Englishes to employ the complementiser that more readily than Inner-Circle Englishes. Still, complementising that occurs only in 12 out of the 622 (1.93%) quotative examples analysed—more specifically 7 times with TELL, 4 times with SAY, and 1 time with ASK. Although complementising that may generally be considered absent from Inner-Circle Englishes, its contexts and frequencies of use also appear relatively limited in SLE. In this light, the analysis of complementising that is insightful, but this factor could not be meaningfully integrated into the multifactorial analysis of BE like versus SAY either.
A characteristic that—at least in terms of its formal realisation—is probably variety-specific in SLE is the occurrence of kiyala marking “the end of a quotation, reported statement, indirect question, etc.” (Meyler, 2007, p. 139). The use of kiyala in SLE is a product of language contact between Sinhala and English in Sri Lanka. For Sinhala, Gair (1970, p. 41) describes that kiyala “marks a preceding form as a thought, supposition, quotation, hypostasis, or attributed name” and occurs directly after said material in the clause. An example of kiyala in Sinhalese, which is taken from Ananda (2011, p. 84), is provided in (18). Here, kiyala immediately follows quoted spoken material, which is thus in line with Gair’s (1970, p. 41) profile of kiyala, but Ananda (2011, p. 84) adds the pragmatic observation that “[t]he speaker is merely reporting the event for whose truth he/she is not ready to undertake any commitment. […] The real interpretive content of this is that it is exactly not certain whether Ravi washed the car: Nimal reports so and may be Ravi actually did not wash the car. This establishes the fact that kiyala is the true quotative complementizer in Sinhala.” Kiyala can also be used to mark English quoted material when the matrix language is Sinhala, instantiating instances of Sinhala-to-English code switching (Wickramasinghe, 2024). In (19), which has been adopted from Wickramasinghe (2024, p. 20), the English quote you look gorgeous is followed by kiyala and otherwise embedded in Sinhala text to mean “Do you know he said that I am gorgeous?”.
(18)
Nimal - kiuwa - [Ravi- ka:reka - seeduwa - kiyala]
N(NOM) - said - [Ravi(NOM) - car - washed - COMP]
‘Nimal said that Ravi washed the car.’ (Ananda, 2011, p. 84)
(19)
oya dannawada eya kiwwa ‘you look gorgeous’ kiyala (Wickramasinghe, 2024, p. 20)
When we now analyse the SLE material at hand, it becomes apparent that the use of the quotative particle kiyala in SLE is grammatically aligned with that in Sinhala in that it is also placed to the immediate right of quoted material, as evident from (20) and (21).
(20)
<ICE-SL:S1A-019#149:1:A> So I thought I’ll join the next <,> round kiyala
(21)
<ICE-SL:S1A-022#166:1:A> So they were like do that if you want kiyala
Kiyala occurs four times with BE like, three times with TELL and once with ASK, THINK, and SAY. While the contents and contexts of the quoted material in said examples do not allow us to comment on whether the pragmatic feature of speakers removing themselves from the truthfulness of what is quoted has also been carried over from Sinhala into SLE, the examples concerned highlight the sociolinguistics of kiyala. Female and male speakers in their early twenties with different occupations employ kiyala. The fact that an ethnically Tamil speaker who profiles English as her home language uses the originally Sinhala kiyala may indicate that this feature has become an integral part of the SLE language system and is not in an ad hoc fashion transferred into English discourse by speakers whose first language is Sinhala. Yet, given that kiyala overall only occurred rarely with BE like or SAY, it was not considered in their multifactorial comparison.

4.2. Quotatives BE like vs. SAY in Sri Lankan English

As indicated in Section 3.3, the choice between the two variants that jointly account for 4 out 5 quotatives in SLE, i.e., BE like and SAY, is modelled via a glmertree. Out of the 500 cases of BE like and SAY, only 411 could be retained for multifactorial analysis because they featured complete sets of metainformation.
The glmertree predicts the choice between BE like and SAY as the dependent variable with the help of the 15 variables documented and illustrated in Section 3.2, which function as independent variables. As explained in Section 3.3, the set of independent variables needed to be and was extended through binary interaction predictors between all independent variables. To form meaningful interaction predictors with numeric variables such as age or quote_length, numeric predictors were transformed into binary categorical predicators through the creation of two levels: one level for values above the mean of the predictor and one for values equal to or below the mean. Accounting for repeated measures, i.e., several quotatives produced by the same speaker, the glmertree features speaker as a random effect. The resulting glmertree has a C score of 0.93 and a classification accuracy of 84.43%, which is statistically highly significantly (p < 0.001) better than a baseline model always predicting the more frequent quotative choice, namely BE like, which would reach a classification accuracy of 63.75%. Figure 1 illustrates the speaker-specific variability captured through the random effect for speaker and Figure 2 shows the glmertree for the choice between BE like and SAY in SLE.
Positive values for conditional variances in Figure 1 display a speaker-specific tendency towards SAY, negative values towards BE like. Speaker A in the first subtext of the file S1A-011 at the very top of Figure 1, a 19-year-old female secondary-school graduate with a Sinhalese background and English as her home language, employs SAY five times in the data without using BE like once. While elderly speakers of 60+ years also demonstrate strong tendencies towards SAY, there are also a number of younger speakers who (almost) exclusively opt for SAY when they could have chosen BE like. In contrast, speaker S1A-068-1:A at the bottom of Figure 1 is a 41-year-old male Sri Lankan Moor who works as an event manager and DJ listing English as his home language. He—despite his seemingly atypical sociobiographic profile—prefers BE like over SAY.
The glmertree in Figure 2 shows that out of the 15 structural and sociobiographic predictors as well as their interaction predictors only 4, i.e., age, year_of_recording, and home language in interaction with tense, significantly influence the choice between BE like and SAY in SLE. The most important split in the data relates to age in node 1, separating speakers with a logged age of larger than 3.332, i.e., older than 28 years, with a marked preference for SAY from the remaining speakers. year_of_recording splits the data up further, in that speakers aged 28 or younger have an on average stronger inclination towards BE like from 2016 onwards than before. In the period from 2012 to 2015, the interaction predictor home_language_tense matters for speakers who are 28 or younger, in that BE like is particularly prominent in the historical present across both English and Sinhala as home languages and non-past contexts with Sinhala as the language of the home. This is exemplified in (22), where a then 23-year-old woman with Sinhala as her home language uses BE like in historical present tense during a recording made in 2012. In the remaining constellations of home language and tense, SAY is notably more frequent, although BE like is still preferred. One such case of quotative SAY in past tense is shown in (23), produced in 2013 by a then 23-year-old woman with Sinhala as her home language.
(22)
<$B><ICE-SL:S1A-016#325:1:B> It was <,> okay it was just after Jaff I mean during camp no
<$A><ICE-SL:S1A-016#326:1:A> Ahh they would have met up afterwards
<$B><ICE-SL:S1A-016#327:1:B> Yeah they met <,> April eighteenth they met and by that time nangi also told Arshada
<ICE-SL:S1A-016#328:1:B> Then Arshada is like I can’t wait without telling when we meet
(23)
<ICE-SL:S1A-019#228:1:B> I told him <,> if you want go for classes and I said <,> even if I ask him do it in Sinhala I don’t have Sinhala notes no

5. Discussion and Conclusions

This discussion focuses on the factors that have been profiled as statistically relevant for the choice of the quotative forms BE like vs. SAY in SLE. Still, this section will also look into some of the factors that do not appear relevant for said choice based on the statistical modelling process, but have been described as relevant for quotatives in earlier research. Finally, the facets of localisation of the SLE quotative system will be discussed.
The most decisive factor for the choice between BE like and SAY in SLE is speaker age. While Blyth et al. (1990, p. 219) describe BE like as a feature of “teenagers and college-age speakers”, Tagliamonte and D’Arcy (2004, p. 510) put forward that “[f]or now, the question of where be like stops is uncertain”. For SLE speakers, it seems possible to state at least where the tendency towards BE like stops—if SLE speakers are 28 years of age or younger, they tend towards BE like, if not, they tend towards SAY. This contrasts with Davydova’s (2016, p. 190) observations for IndE that “say is favoured in the speech of younger (18–25) and older (>42) male speakers”, which may be indicative of (a) cross-varietal quotative differences in South Asian Englishes or, maybe more likely, (b) a diachronic integration of BE like into South Asian Englishes, given that the Indian English data in her analysis stem from the Indian ICE component featuring texts from the early 1990s. Another reason why interpretation (b) appears sociolinguistically more adequate is that BE like also tends to be associated with younger speakers in other Outer-Circle varieties like Ghanaian English, as observed in contemporary corpus data from the Ghanaian ICE component (Hansen, 2021, p. 39). This connection between a preference for BE like and younger speaker groups that are probably not yet under maximum social pressure to conform to overt prestige norms in their respective speech communities may imply that BE like currently functions as a cross-varietal marker of youth identity. In the years to come, it will be relevant to understand whether the younger speakers studied here will (a) drop BE like in favour of SAY as they grow older and social pressure to use traditional standard variants builds, (b) keep BE like as a prominent variant in their linguistic repertoire also through adulthood, potentially triggering a language change away from quotative SAY, or (c) remain associated with BE like as their youth marker while the next generation of younger speakers may craft its own quotative competitor to SAY.
The effect of the predictor year_of_recording indicates the point in time when BE like became particularly prominent in the SLE quotative system. With the younger SLE speaker group, BE like was already the dominant quotative form in conversations that happened in 2015 or earlier, but from 2016 onwards, BE like becomes even more prevalent. In her study on Irish English, Höhn (2012, p. 286) observes a maybe slightly less radical spread of BE like when she states that “unsurprisingly, the use of be like is favoured in the latest collection period”.
In terms of the interaction between home_language and tense, earlier research only relates to tense. Davydova (2019, p. 126) finds BE like to be regularly triggered by the historical present and to a certain extent past tense in IndE. For Ghanaian English, Hansen (2021, p. 39) finds a similar constellation for the occurrence of BE like. This is no different for the SLE speech community—BE like compared to SAY is most dominant with the historical present independent of the home_language the speakers specified when younger speakers talked to one another earlier than 2016. In contrast, in comparative past-tense settings, SAY is employed almost as often as BE like, although there is variability in that non-past tense triggers more uses of SAY with SLE speakers using English at home than with SLE speakers whose home language is Sinhala.
Most of the relevant predictors and their effects on BE like versus SAY may not come as a surprise in the light of the earlier findings outlined in Section 2.1. Still, it is also relevant to explore (a) which factors discussed in previous quotative research on World Englishes appeared influential, but did not turn out to be for the choice between BE like and SAY in SLE and (b) what implications this may have.
The content of the quote has been examined and discussed in almost all studies on BE like and SAY in World Englishes. While SAY prototypically introduces re-created speech, BE like seems to have entered American English through the introduction of inner monologue and, over time, BE like expanded its function to also introduce quoted speech (Ferrara & Bell, 1995, p. 279; Tagliamonte & D’Arcy, 2004, p. 206). The data show that both BE like and SAY can be used to introduce imagined or spoken content. BE like is used more often than SAY in both cases, which is particularly pronounced with imagined content. This goes to show that BE like has extended its functional range in SLE over time or that it was introduced from other varieties into SLE when it had already expanded its use to also introducing speech. Contemporarily, the discriminatory power of this predictor is not strong enough to be a significant one for the choice between BE like and SAY in SLE.
In terms of the person of the subject, a similar process of expansion has been described in the literature for BE like. The use of BE like is at an advanced stage in a given variety if BE like is used with third-person subjects and not only with first-person ones (Ferrara & Bell, 1995). In the data at hand, BE like is the dominant choice across all potential persons of the subject, which does not mark a statistically significant contrast with SAY.
Ferrara and Bell (1995, p. 285) argue for a similar levelling effect with regard to speaker gender in that the spread of BE like in a variety’s quotative system will eliminate gender differences in its use, although Tagliamonte and D’Arcy (2004, 2007) find a strengthening of the gender effect. In the SLE dataset, BE like is more often used by women than by men while male speakers slightly prefer SAY. Still, this difference did not turn out to be significant when other, more important predictors are considered. The absence of statistically significant effects for the content of the quote, the person of the grammatical subject, and speaker gender as regards the choice between BE like and SAY suggests a strong entrenchment of BE like in the SLE quotative system.
Further, none of the predictors that this study explored for the first time with regard to the choice between BE like and SAY displayed a significant discriminatory power. Structurally, the length of the quoted material did not matter for the choice between BE like and SAY. In terms of the sociobiographic predictors, it did not matter whether Sri Lankan speakers had spent a stay abroad or not and also the question whether they consumed (more) American or British media did not make a statistically significant difference. Still, it might have been expected that speakers with a stronger inclination towards American English—either through stays abroad or media exposure—might adopt BE like more readily than speakers with a preference for British English. Similarly, the ethnic group Sri Lankan speakers identified with was not decisive for quotative choices, although ethnicity correlates with home_language featured in a significant interaction predictor with tense.
To extend the scope of the discussion beyond BE like and SAY, D’Arcy (2013, p. 497) reports on the basis of ICE data from the 1990s that the Outer-Circle quotative “repertoires are more constrained than those of the Inner Circle in that fewer verbs are used to introduce direct quotation”. The quotatives D’Arcy (2013, p. 497) finds are ASK, BE like, GO, SAY, TELL, THINK, and zero quotatives. Section 4.1 shows that these quotatives are also used in contemporary SLE, but that speakers also frequently choose additional verbs to introduce quoted material. Although the quotative verbs are certainly not variety-specific as such, other structural features indicate the localisation of the SLE quotative system. The most overt marker of this localisation is probably the use of kiyala (see examples (20) and (21)), a structure transferred from Sinhala and now an integral part of SLE, to mark the end of a quotation. Together with the occasional use of complementising that preceding quoted material (see examples (16) and (17)), a characteristic largely absent from Inner-Circle Englishes (D’Arcy, 2013, p. 498), the SLE quotative system—though maybe not at its core, but rather at its periphery—may be considered nativised.
In sum, this study empirically explored the quotative system of a relatively informal kind of spoken SLE. It became obvious that—in these relatively relaxed communicative settings—BE like is the dominant quotative choice, followed by SAY. While a large number of other quotatives are also used in spoken SLE, their frequencies are generally lower. The choice between BE like and SAY in informal spoken SLE is guided by the age of the speaker, the time at which a quotative was used, and the interaction between home_language and tense. BE like is chosen particularly frequently by speakers who are 28 years or younger when (a) the quotative was produced from 2016 onwards or—in case this does not apply—(b) historical present is used. From various angles, BE like dominates the quotative system of SLE. In comparison to the frequencies of BE like reported for the only other South Asian English that has been studied with regard to quotatives, namely Indian English, the rates of BE like are notably higher in SLE than in Indian English as reported in Davydova (2016, 2019, 2021). There are conceptual and methodological avenues of reasoning as to where these differences may come from.
Conceptually, it is conceivable that certain social networks drive the use of BE like in SLE. The face-to-face conversations in ICE-SL analysed here represent sections of larger social networks since SLE speakers have not been grouped together randomly for the recordings of SLE conversations, which would also go against the corpus-linguistic principle of authenticity, but the face-to-face conversations are by default informal conversations between members of the same social network, i.e., people who intimately know each other like close friends or family. With a focus on these individual conversations representing parts of larger social networks, some are dominated by BE like while others do not feature BE like at all or only rarely. In the conversations/sections of social networks where BE like is most dominant, the users of BE like are female speakers in their 20s. Consequently, potential social network effects as they also manifest themselves in other postcolonial Englishes (cf. Botha & Bernaisch, 2025) could be scrutinised more carefully in future work on quotatives in SLE and other World Englishes.
There might also be a methodological reason for the differences in BE like frequencies reported for Indian English and SLE. In comparison to more socially neutral quotatives, BE like is associated with positive, but also with negative connotations, the latter of which ESL and EFL speakers are also aware of (Davydova, 2019). The Indian English data for quotative analyses in Davydova (2016, 2019, 2021) stem from the private and public conversations in ICE India and recent sociolinguistic interview data whereas the Sri Lankan data have exclusively been sourced from private face-to-face conversations. ICE-India’s original sample period ranges from 1990 to 1994, a time when BE like had only recently been documented (Butters, 1982) and had probably not yet entered quotative systems to the degree it has today. Further, the inclusion of texts recorded in public in the Indian, but not the Sri Lankan dataset studied here, such as classroom lessons, also has to be considered, because this inclusion introduces pressures on speakers to conform to overt linguistic prestige norms that might discourage the use BE like in sections of the Indian data, which are, however, absent from the Sri Lankan data analysed. Consequently, the earlier recording dates along with the higher degree of formality/public pressure represented in the Indian ICE data might contribute to the lower frequencies of BE like. Similarly, sociolinguistic interviews conducted by a linguistic expert might exert comparable linguistic pressures to conform to overt linguistic standards and lead participants to less frequently opt for BE like in comparison to the informal and unsupervised face-to-face conversations in ICE-SL.
Various future avenues for research emerge from this study. The scope of this paper has been limited to SLE, although it might be desirable to statistically compare quotatives systems across different regional varieties—be they part of the Inner, Outer, or Expanding Circle. Still, it was the intention of this project to provide a thick sociolinguistic description of the SLE quotative system through the integration of as much sociobiographic information as possible. If, for example, another ICE component had been featured for a cross-varietal comparison, this would have severely limited the number of different sociobiographic factors that could have been modelled multifactorially, since many ICE components do not feature the same depth of sociobiographic information.
The exploration of the sensitivity of quotative choices at different levels of formality and in speech versus writing would probably prove insightful as well. It would be highly surprising to find BE like to be as dominant in, say, legal proceedings as in face-to-face conversations. Similarly, BE like can also be expected to be less frequent in the written than in the spoken mode. Still, metainformation on speakers and writers in ICE-SL is notably more limited in corpus sections other than face-to-face conversations. Consequently, restrictions in terms of sociobiographic details similar to those speaking against the integration of another ICE component hindered the inclusion of texts from written and other spoken genres represented in ICE-SL.
The face-to-face conversations in ICE-SL were used to provide short-term diachronic insights into the development of the quotative system in SLE. While this dataset features carefully transcribed texts and rich sociobiographic annotation, it certainly has limitations in terms of overall size, which might have potentially masked the effects of predictors for quotatives evident from earlier research documented in Table 1. It would thus be highly desirable to collect more data of the sort analysed here or to go back further in time to trace the emergence of BE like and how its integration into SLE in the face of other competing quotative forms proceeded over time. Thus, the compilation of more (ideally richly sociobiographically annotated) synchronic as well as diachronic corpora of SLE would be most welcome.

Funding

This research received no external funding.

Institutional Review Board Statement

This research analyses data from the International Corpus of English, so it is exempt from Ethical Committees/Institutional Review Boards’ review and statement.

Informed Consent Statement

Not applicable.

Data Availability Statement

The original contributions presented in this study are included in the article. Further inquiries can be directed to the corresponding author.

Conflicts of Interest

The authors declare no conflicts of interest.

Abbreviations

The following abbreviations are used in this manuscript:
EFLEnglish as a Foreign Language
ESLEnglish as a Second Language
ICE-SLThe Sri Lankan Component of the International Corpus of English
SAVEThe South Asian Varieties of English Corpus
SAVE2020The 2020 update of the South Asian Varieties of English Corpus
SLESri Lankan English

Notes

1
For comprehensive overviews of research on SLE, please consult Meyler (2013) or Ekanayaka (2020). For detailed descriptions of the history of (English in) Sri Lanka, please refer to de Silva (1981), Yogasundram (2008) or Coperahewa (2009).
2
Please note the critical discussion of GloWbE in the World Englishes community—also with regard to the Sri Lankan data available there (Mukherjee, 2015).
3
Earlier research on quotatives in World Englishes as presented in Section 2.1 has habitually contrasted the use of BE like with all other quotative forms found in the datasets under scrutiny. This paper departs from this approach in that it explicitly zooms in on the choice between the quotatives BE like and SAY in SLE since (a) they together constitute the vast majority of quotative forms in SLE (80.39%) and (b) SAY—with a relative frequency of 28.14%—appears too frequently in comparison to the other rarer quotative forms, which individually are all below relative frequencies of 7%, to group them together because it would mask the dominance of and the primary applicability of results to SAY in this group.

References

  1. Ananda, M. G. L. (2011). Complementizer distribution in Sinhala. Vidyodaya Journal of Humanities and Social Sciences, 3, 73–92. [Google Scholar] [CrossRef]
  2. Barbieri, F. (2005). Quotative use in American English: A corpus-based, cross-register comparison. Journal of English Linguistics, 33(3), 222–256. [Google Scholar] [CrossRef]
  3. Barbieri, F. (2009). Quotative be like in American English: Ephemeral or here to stay? English World-Wide, 30(1), 68–90. [Google Scholar] [CrossRef]
  4. Bernaisch, T. (2015). The lexis and lexicogrammar of Sri Lankan English. John Benjamins. [Google Scholar] [CrossRef]
  5. Bernaisch, T. (2022). Comparing generalised linear mixed-effects models, generalised linear mixed-effects model trees and random forests: Filled and unfilled pauses in varieties of English. In O. Schützler, & J. Schlüter (Eds.), Data and methods in corpus linguistics: Comparative approaches (pp. 163–193). Cambridge University Press. [Google Scholar]
  6. Bernaisch, T. (in press). Sri Lankan English. In R. Hickey (Ed.), The new Cambridge history of the English language. Cambridge University Press.
  7. Bernaisch, T., Heller, B., & Mukherjee, J. (2021). Manual for the 2020-update of the South Asian varieties of English (SAVE2020) corpus. Version 1.1. Department of English, Justus Liebig University. [Google Scholar]
  8. Bernaisch, T., Koch, C., Mukherjee, J., & Schilk, M. (2011). Manual for the South Asian varieties of English (SAVE) corpus: Compilation, cleanup process, and details on the individual components. Justus Liebig University, Department of English. [Google Scholar]
  9. Bernaisch, T., Mendis, D., & Mukherjee, J. (2019). Manual to the international corpus of English—Sri Lanka. Department of English, Justus Liebig University. [Google Scholar]
  10. Blyth, C. J., Recktenwald, S., & Wang, J. (1990). I’m like, “Say what?!”: A new quotative in American oral narrative. American Speech, 65(3), 215–227. [Google Scholar] [CrossRef]
  11. Bogetić, K. (2014). Be like and the quotative system of Jamaican English: Linguistic trajectories of globalization and localization. English Today, 30(3), 5–12. [Google Scholar] [CrossRef]
  12. Botha, W., & Bernaisch, T. (2025). Social network effects on particle variation among Singapore students. World Englishes, 44(1–2), 144–165. [Google Scholar] [CrossRef]
  13. Buchstaller, I. (2006). Diagnostics of age-graded linguistic behaviour: The case of the quotative system. Journal of Sociolinguistics, 10, 3–30. [Google Scholar] [CrossRef]
  14. Buchstaller, I. (2008). The localization of global linguistic variants. English World-Wide, 29(1), 15–44. [Google Scholar] [CrossRef]
  15. Buchstaller, I., & D’Arcy, A. (2009). Localized globalization: A multi-local, multivariate investigation of quotative be like. Journal of Sociolinguistics, 13(3), 291–331. [Google Scholar] [CrossRef]
  16. Buchstaller, I., Rickford, J., Traugott, E., Wasow, T., & Zwicky, J. (2010). The sociolinguistics of a short-lived innovation: Tracing the development of quotative all across spoken and internet newsgroup data. Language Variation and Change, 22, 191–219. [Google Scholar] [CrossRef]
  17. Butters, R. (1980). Narrative go “say”. American Speech, 55, 304–307. [Google Scholar] [CrossRef]
  18. Butters, R. (1982). Editor’s note. American Speech, 57, 149. [Google Scholar] [CrossRef]
  19. Cheshire, J., Kerswill, P., Fox, S., & Torgersen, E. (2011). Contact, the feature pool and the speech community: The emergence of multicultural London English. Journal of Sociolinguistics, 15, 151–196. [Google Scholar] [CrossRef]
  20. Coperahewa, S. (2009). The language planning situation in Sri Lanka. Current Issues in Language Planning, 10(1), 69–150. [Google Scholar] [CrossRef]
  21. D’Arcy, A. (2010). Quoting ethnicity: Constructing dialogue in Aotearoa/New Zealand. Journal of Sociolinguistics, 14(1), 60–88. [Google Scholar] [CrossRef]
  22. D’Arcy, A. (2012). The diachrony of quotation: Evidence from New Zealand English. Language Variation and Change, 24(3), 343–369. [Google Scholar] [CrossRef]
  23. D’Arcy, A. (2013). Variation and change. In R. Bayley, R. Cameron, & C. Lucas (Eds.), The Oxford handbook of sociolinguistics (pp. 484–502). Oxford University Press. [Google Scholar]
  24. D’Arcy, A. (2015). Quotation and advances in understanding syntactic systems. Annual Review of Linguistics, 1, 43–61. [Google Scholar] [CrossRef]
  25. Davies, M. (2013). Corpus of global web-based English. Available online: https://www.english-corpora.org/glowbe/ (accessed on 20 June 2025).
  26. Davies, M. (2016). Corpus of news on the web (NOW). Available online: https://www.english-corpora.org/now/ (accessed on 20 June 2025).
  27. Davydova, J. (2016). Indian English quotatives in a diachronic perspective. In E. Seoane, & C. Suárez-Gómez (Eds.), World Englishes: New theoretical and methodological considerations (pp. 173–204). John Benjamins. [Google Scholar] [CrossRef]
  28. Davydova, J. (2019). Quotation in indigenised and learner English: A sociolinguistic account of variation. De Gruyter Mouton. [Google Scholar] [CrossRef]
  29. Davydova, J. (2021). The role of sociocognitive salience in the L2 acquisition of structured variation and linguistic diffusion: Evidence from quotative be like. Language in Society, 50(2), 171–196. [Google Scholar] [CrossRef]
  30. Degenhardt, J. (2025). Parentheticals in spoken Indian and Sri Lankan English. World Englishes, 44(1–2), 184–201. [Google Scholar] [CrossRef]
  31. de Silva, K. M. (1981). A history of Sri Lanka. University of California Press. [Google Scholar]
  32. Deuber, D., Hänsel, E. C., & Westphal, M. (2021). Quotative be like in Trinidadian English. World Englishes, 40(3), 436–458. [Google Scholar] [CrossRef]
  33. DHLab—Department of English. (2022). The Sri Lanka English newspaper corpus 1.0. University of Colombo. Available online: https://dhlab.cmb.ac.lk/slenc (accessed on 20 June 2025).
  34. Ekanayaka, T. N. I. (2020). Sri Lankan English. In K. Bolton, W. Botha, & A. Kirkpatrick (Eds.), The handbook of Asian Englishes (pp. 337–353). John Wiley & Sons. [Google Scholar] [CrossRef]
  35. Ferrara, K., & Bell, B. (1995). Sociolinguistic variation and discourse function of constructed dialogue introducers: The case of be + like. American Speech, 70(3), 265–290. [Google Scholar] [CrossRef]
  36. Fokkema, M., Smits, N., Zeileis, A., Hothorn, T., & Kelderman, H. (2018). Detecting treatment-subgroup interactions in clustered data with generalized linear mixed-effects model trees. Behavior Research Methods, 50(5), 2016–2034. [Google Scholar] [CrossRef]
  37. Fox, S. (2012). Performed narrative: The pragmatic function of this is + speaker and other quotatives in London adolescent speech. In I. Buchstaller, & I. van Alphen (Eds.), Quotatives: Cross-linguistic and cross-disciplinary perspectives (pp. 213–257). John Benjamins. [Google Scholar] [CrossRef]
  38. Gair, J. W. (1970). Colloquial Sinhalese grammar and clause structure. Mouton. [Google Scholar] [CrossRef]
  39. Greenbaum, S. (1996). Introducing ICE. In S. Greenbaum (Ed.), Comparing English worldwide: The international corpus of English (pp. 3–12). Clarendon. [Google Scholar]
  40. Gries, S. T. (2020). On classification trees and random forests in corpus linguistics: Some words of caution and suggestions for improvement. Corpus Linguistics and Linguistics Theory, 16(3), 617–647. [Google Scholar] [CrossRef]
  41. Gries, S. T., & Bernaisch, T. (2016). Exploring epicentres empirically: Focus on South Asian Englishes. English World-Wide, 37(1), 1–25. [Google Scholar] [CrossRef]
  42. Gunesekera, M. (2005). The postcolonial identity of Sri Lankan English. Katha Publishers. [Google Scholar]
  43. Hansen, B. (2021). Localisation, globalisation and gender in discourse-pragmatic variation in Ghanaian English. In T. Bernaisch (Ed.), Gender in world Englishes (pp. 23–46). Cambridge University Press. [Google Scholar] [CrossRef]
  44. Hothorn, T., & Zeileis, A. (2015). partykit: A modular toolkit for recursive partytioning in R. Journal of Machine Learning Research, 16, 3905–3909. [Google Scholar]
  45. Höhn, N. (2012). “And they were all like ‘What’s going on?’”: New quotatives in Jamaican and Irish English. In M. Hundt, & U. Gut (Eds.), Mapping unity and diversity world-wide: Corpus-based studies of New Englishes (pp. 263–290). John Benjamins. [Google Scholar] [CrossRef]
  46. Huber, M. (1999). Ghanaian Pidgin English in its West African context: A sociohistorical and structural analysis. John Benjamins. [Google Scholar] [CrossRef]
  47. Hundt, M., Dallas, B., & Nakanishi, S. (2024). The be- versus get-passive alternation in world Englishes. World Englishes, 43(1), 86–108. [Google Scholar] [CrossRef]
  48. Kortmann, B., Lunkenheimer, K., & Ehret, K. (Eds.). (2020). The electronic world atlas of varieties of English. Zenodo. Available online: http://ewave-atlas.org (accessed on 21 August 2025). [CrossRef]
  49. Meyerhoff, M. (2003). Formal and cultural constraints on optional objects in Bislama. Language Variation and Change, 14, 323–346. [Google Scholar] [CrossRef]
  50. Meyler, M. (2007). A dictionary of Sri Lankan English. Mirisgala. [Google Scholar]
  51. Meyler, M. (2013). Sri Lankan English. In B. Kortmann, & K. Lunkenheimer (Eds.), The mouton world atlas of variation in English (pp. 540–547). De Gruyter. [Google Scholar] [CrossRef]
  52. Mukherjee, J. (2015). Response to mark Davies and Robert Fuchs: Expanding horizons in the study of World Englishes with the 1.9 billion word global web-based English corpus (GloWbE). English World-Wide, 36(1), 34–37. [Google Scholar] [CrossRef]
  53. Nelson, G. (1996). The design of the corpus. In S. Greenbaum (Ed.), Comparing English worldwide: The international corpus of English (pp. 27–35). Clarendon. [Google Scholar]
  54. Rautionaho, P., & Hundt, M. (2021). Primed progressives? Predicting aspectual choice in World Englishes. Corpus Linguistics and Linguistic Theory, 18(3), 599–625. [Google Scholar] [CrossRef]
  55. R Core Team. (2024). R: A language and environment for statistical computing. R Foundation for Statistical Computing. [Google Scholar]
  56. Rickford, J. R., Wasow, T., Zwicky, A., & Buchstaller, I. (2007). Intensive and quotative all: Something old, something new. American Speech, 83, 3–31. [Google Scholar] [CrossRef]
  57. Schneider, E. W. (2007). Postcolonial English: Varieties around the world. Cambridge University Press. [Google Scholar]
  58. Senaratne, C. D. (2009). Sinhala-English code-mixing in Sri Lanka: A sociolinguistic study. LOT. [Google Scholar]
  59. Singler, J. V. (2001). Why you can’t do a VARBRUL study of quotatives and what such a study can show us. University of Pennsylvania Working Papers in Linguistics, 7(3), 257–278. [Google Scholar]
  60. Tagliamonte, S., & D’Arcy, A. (2004). He’s like, she’s like: The quotative system in Canadian youth. Journal of Sociolinguistics, 8(4), 493–514. [Google Scholar] [CrossRef]
  61. Tagliamonte, S., & D’Arcy, A. (2007). Frequency and variation in the community grammar: Tracking a new change through the generations. Language Variation and Change, 19(2), 199–217. [Google Scholar] [CrossRef]
  62. Tagliamonte, S., & Hudson, R. (1999). Be like et al. beyond America: The quotative system in British and Canadian youth. Journal of Sociolinguistics, 3(2), 147–172. [Google Scholar] [CrossRef]
  63. Wickramasinghe, S. T. A. (2024). Sinhala-English code-switching in text messaging: A study based on undergraduates of two state universities in Sri Lanka. In 5th SLIIT International Conference on Advancements in Sciences and Humanities (pp. 196–201). Sri Lanka Institute of Information Technology, Faculty of Humanities and Sciences. [Google Scholar]
  64. Xiao, R. (2009). Multidimensional analysis and the study of world Englishes. World Englishes, 28(4), 421–450. [Google Scholar] [CrossRef]
  65. Yogasundram, N. (2008). A comprehensive history of Sri Lanka: From prehistory to tsunami (2nd ed.). Vijitha Yapa Publications. [Google Scholar]
Figure 1. The random effect for speaker in the glmertree.
Figure 1. The random effect for speaker in the glmertree.
Languages 10 00236 g001
Figure 2. Glmertree for the choice between BE like and SAY.
Figure 2. Glmertree for the choice between BE like and SAY.
Languages 10 00236 g002
Table 1. Predictors in earlier World Englishes research on quotatives. Ticks show that a predictor has been studied and pluses that a predictor has been reported as statistically significant.
Table 1. Predictors in earlier World Englishes research on quotatives. Ticks show that a predictor has been studied and pluses that a predictor has been reported as statistically significant.
Höhn (2012) on BE like and SAY in Irish and
Jamaican English
D’Arcy (2013) on BE like in Hong Kong, Kenyan, Philippine, and
Singapore English
Bogetić (2014) on BE like in
Jamaican English
Davydova (2016) on BE like and SAY in Indian EnglishDavydova (2019) on BE like in Indian EnglishDavydova (2021) on BE like in Indian English and English in GermanyDeuber et al. (2021) on BE like in
Trinidadian English
Hansen (2021) on BE like in
Ghanaian English
STRUCTURAL PREDICTORS
Mimesis ✓+✓+
Person of grammatical subject✓+✓+✓+✓+
Quoted material✓+✓+✓+✓+✓+ ✓+
Tense ✓+✓+✓+✓+ ✓+
SOCIOBIOGRAPHIC PREDICTORS
Age ✓+ ✓+✓+
Education ✓+
Gender✓+ ✓+ ✓+✓+
Language context
Media exposure
Occupation ✓+
Region
Time✓+
Table 2. Absolute frequencies and proportions of quotatives in the face-to-face conversations of ICE-SL.
Table 2. Absolute frequencies and proportions of quotatives in the face-to-face conversations of ICE-SL.
QuotativeAbsolute FrequencyProportion
BE like32552.25%
SAY17528.14%
TELL386.11%
ASK325.14%
THINK233.7%
zero142.25%
GO40.64%
Other111.77%
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Share and Cite

MDPI and ACS Style

Bernaisch, T. The Sociolinguistics of Quotatives in Sri Lankan English: Corpus-Based Insights. Languages 2025, 10, 236. https://doi.org/10.3390/languages10090236

AMA Style

Bernaisch T. The Sociolinguistics of Quotatives in Sri Lankan English: Corpus-Based Insights. Languages. 2025; 10(9):236. https://doi.org/10.3390/languages10090236

Chicago/Turabian Style

Bernaisch, Tobias. 2025. "The Sociolinguistics of Quotatives in Sri Lankan English: Corpus-Based Insights" Languages 10, no. 9: 236. https://doi.org/10.3390/languages10090236

APA Style

Bernaisch, T. (2025). The Sociolinguistics of Quotatives in Sri Lankan English: Corpus-Based Insights. Languages, 10(9), 236. https://doi.org/10.3390/languages10090236

Article Metrics

Back to TopTop