With perceptual dialectology—a sub-area of study in folk linguistics—we explore non-linguists’ perceptions of where linguistic variation exists. One of the main ways this is accomplished involves asking participants to draw lines on a map in order to delimit where they believe varieties are different from one another in a specific region (Preston 2013; 1981). This is the so-called ‘draw-a-map’ task which has been utilized in a number of studies all around the world (Al-Rojaie 2020; Bounds 2015; Bucholtz et al., 2007; Cramer 2021; Evans 2013b; Evans, Dunbar, and Chartier 2020; Garrett, Coupland, and Williams 2003; Long and Preston 2002; Montgomery 2007; Preston, 1989; Preston, 1999; Theodoropoulou and Tyler 2014). In addition to the map task, some studies have also analyzed data from interviews (e.g., Montgomery, 2014). Data from perceptual dialectology studies are valuable because they can uncover certain linguistic patterns, stereotypes, and attitudes that speakers have toward the speech of a particular region (Preston 2004; cited in Evans, 2013b, 282). They also add to our understanding of bigger questions about the forces behind language variation and change (Gordon 2013, 227) and contribute to data triangulation in studies of language variation and change by combining perception with production data (Meyerhoff 2016, 442). As Cramer (2021) aptly points out, perceptual dialectology “serves as a framework for connecting perception and production, attempting to present a fuller view of the complex nature of linguistic variation” (2021, 2). Such studies can illuminate which linguistic features are salient to speakers and which ones they tend to disregard, and whether or not their perceptions match data from production studies (Gordon 2013; Theodoropoulou and Tyler 2014).

This paper contributes to the literature in two ways. To start with, it presents the results of the first perceptual dialectological analysis of linguistic variation in Greek-speaking Cyprus. While there are studies on language attitudes in the community (e.g., Ayiomamitou and Fotiou 2021; Fotiou and Ayiomamitou 2021; Papapavlou 1998), there are no studies adopting a perceptual dialectology approach to date in this context. Second, this study adds to a few recent studies in the literature (e.g., Evans 2013b) that investigate perceptions on linguistic variation in small places. Cyprus, which is an island situated in the Eastern Mediterranean, covers an area of just 9,251 km² (see Figure 1 for a map of Cyprus). Most studies on perceptual dialectology investigate people’s perception of regional variation in very large geographical areas. For example, Theodoropoulou and Tyler (2014), asked their participants to draw a map of the Arab world—which they define as the countries of the Arab League—while Evans et al. (2020) asked residents of Cardiff about their perceptions of English in the United Kingdom. In a similar vein, Long (1999) investigated people’s perception of regional variation in Japan and Preston (1981, 2013) focused on people’s perception of regional variation in the United States. Understandably, the study of people’s perception of geographical variation in relation to large geographical areas such as Japan, the United States and the United Kingdom is likely to result in not very detailed accounts of regional variation due to the size of the place under study. In contrast, investigating people’s perceptions of regional variation at a smaller scale can provide, as Evans (2013b, 269) argues, a more comprehensive and nuanced picture.

In what follows, we set the scene of this study by describing the sociolinguistic situation of the Greek-speaking community of Cyprus with a focus on the research conducted so far in this setting on regional linguistic variation (section 2). We then describe the methodology we adopted in this study (section 3) before we provide the results of our study (section 4) and discuss them (section 5). Finally, we conclude the paper and offer suggestions for future research (section 6).

The Greek-speaking community of Cyprus has been characterized as a diglossic setting (Rowe and Grohmann 2013). Standard Modern Greek (henceforth SMG) is the sociolinguistically High variety in this setting: It is one of the two official languages of the Republic of Cyprus (the other one being Turkish), it is acquired at school, and it is the variety people (aspire to) use in formal written and oral communication. Cypriot Greek (henceforth CG), which is the mother tongue of Greek Cypriots, is the Low variety, and is used in informal interactions. However, over the past two decades CG has found its way into domains “in which its presence was perhaps unconceivable in the past” (Fotiou and Ayiomamitou 2021, 3), such as the media, the arts (literature, theatre, cinema) and the linguistic landscape of the island. What is more, in spite of official policies, many studies have demonstrated that CG is also used in all levels of education (pre-primary to tertiary) along with SMG (Tsiplakou 2007; Ioannidou 2009; Sophocleous and Ioannidou, 2020).

Older studies of CG conducted in the 1950s and 1960s argued for the presence of geographical dialect zones in Cyprus. For example, Contosopoulos (1969, 105) claimed the existence of eighteen dialect zones, as can be seen on the map in Figure 2. In another study, Newton (1972) provided a detailed description of phonetic variants from 128 villages from around the island and argued for the existence of four main linguistic regions: Central Cyprus, North Paphos, South Paphos and Karpasia. Newton explains that his work was based on a variety of CG which he calls a ‘village dialect’ and that his work was mainly focused on the language of the inhabitants of Mesaoria, the island’s central plain. He also notes that ‘village CG’ is not a homogeneous variety but “a continuum of closely related types of speech linked by a series of independent phonological, morphological and lexical isoglosses (i.e., no obvious ‘bundles’ of isoglosses were found, with the possible exception of those setting off the speech of northern Paphos)” (p. 19). At the same time, he notes the existence of a variety of CG which calls ‘town speech’ (ibid.), and at a later point, he also refers in passing to a ‘local koine’:

Even though Newton (1972) does not say anything else in his work about the CG koine, one may deduce from what he claims that at the time of his writing variation in CG was not just geographical. On the contrary, it seems that “geographical and sociolinguistic variation were both at play, and variants from the same pool may well have functioned as indices of both geographical provenance and of sociolinguistic status” (Tsiplakou et al., 2006, 10).

Today, regional variation on the island is mainly noted in segmental and suprasegmental phonology (Terkourafi 2005, 326), and there has been a shift from a geographical dialect continuum to a register continuum (Papapavlou and Sophocleous, 2009; Tsiplakou, 2014; Tsiplakou and Armostis, 2020). CG is also undergoing levelling of marked local basilect features and subvarieties (Kerswill 2003) and a CG koine is developing (Terkourafi 2005; Karyolemou 2008; Tsiplakou and Kontoyiorgi 2016; Karmellou 2017; Tsiplakou and Armostis 2020). At the same time, features of older local basilects that are still used post-levelling appear to operate as indexicals of lower registers (Papapavlou and Sophocleous, 2009; Tsiplakou et al., 2006, cited in Tsiplakou and Armostis, 2020).

The koine has been formed based on the language of the Mesaoria region (Newton, 1972, 21), which is the island’s central plain, but other varieties have also contributed to its formation (Karyolemou, 2008, p. 451). This koine is sometimes referred to as an urban variety (Davy, Ioannou, and Panayotou 1996; Terkourafi 2005; Savva 2016), but Karyolemou (2008) clarifies that this is so in order to highlight the fact that it was first developed in urban centers. As she explains, it is now spreading towards rural areas, and it has ceased to be a characteristic of urban space. Nevertheless, with only a handful of studies on the speech of people living in rural areas (Christodoulou 2015; Tsiplakou and Kontoyiorgi 2016; Karmellou 2017), we are not aware of the extent of this spread.

A pivotal factor contributing to the formation of the koine was the Turkish invasion of 1974 because it resulted in the forced movement of one third of the Greek Cypriot population from the northern part of the island to the south and the subsequent de facto partition of the island (Terkourafi 2005; Tsiplakou 2014). In fact, as a result of the invasion one can assume that “abrupt levelling must have taken place within the space of one generation, as sudden dense contact among populations from different areas of the island must have induced speakers to shed identifiably local, basilectal variants for the purposes of mutual intelligibility” (Tsiplakou, Armostis, and Evripidou 2016, 11).

Evidently, it was not just the forced population movement that has played its part in the formation of the koine (Terkourafi 2005). Other factors that have contributed to its formation have been easing of transportation, development of telecommunications as well as economic and infrastructural changes (Tsiplakou, Armostis, and Evripidou 2016, 11) combined with people’s development of weak social ties with various (loose-knit) social networks (Terkourafi, 2005; Karyolemou, 2008). As Terkourafi (2005, 330) explains:

By the early 1960s, just over one third of the population lived in urban areas (Karyolemou 2008), while in the latest census of 2011 it was recorded that 67% of Greek Cypriots live in urban areas (CYSTAT 2015, 16). Finally, another factor that has contributed to processes of levelling and koineization in the community is a sharp increase in literacy in the High variety (i.e., SMG) (Hadjioannou, Tsiplakou, and Kappler 2011; Tsiplakou and Kontoyiorgi 2016). The High variety is acquired at school and, in Cyprus, attending primary school became obligatory in 1962, while attending lower secondary education became obligatory in 1985 (Persianis and Polyviou, 1992). As an indication of the sharp increase in literacy, in 1911 73.21% of the population were illiterate (Polidorou 1995), while today 98.7% of the population over the age of 15 is literate (CYSTAT, 2015, 21).

As noted above, there has been a shift from a geographical dialect continuum to a register continuum in this community. As Figure 3 illustrates, the CG koine resides in an intermediate sociolinguistic space in this continuum according to Themistocleous and Tsiplakou (2013, 448)—this register has been characterized by speakers as ‘tidied-up’ or ‘polite’ Cypriot Greek. At one end of this continuum, we have ‘peasantry’, which is also called by speakers ‘heavy Cypriot’ and sometimes just ‘Cypriot’. This is the lowest register which exhibits use of basilectal features. At the other end of the continuum, we find SMG as it is spoken in Cyprus—this variety has been called Cypriot Standard Greek (Arvaniti 2006). It is a variety “with Cypriot phonetic (segmental and suprasegmental) features but (felt to be) identical to Standard Greek in other respects” (Themistocleous and Tsiplakou 2013, 448).

Previous studies have noted that many speakers are aware of the different registers within CG (Tsiplakou et al., 2006; Papapavlou and Sophocleous, 2009) and they employ different emic terms to refer to them. As we will see later in this paper, this observation is confirmed with data from the present study. An example:

Further, many speakers are also aware of the fact that there was regional variation on the island in the past. However, most, and especially those born after 1974, are not able to identify the different regional dialects or note their salient features (Pastella, 2005; cited in Tsiplakou et al., 2006, 26). In fact, while they may argue that they are able to recognize a regional idiom, the only impressionistic criterion they put forward is intonation (Tsiplakou et al., 2006, 26) as well as the fact that some words are pronounced differently in other regional idioms (Terkourafi, 2005, 326). These observations are also confirmed by the data of our study.

While people’s perceptions on regional variation in Cyprus have been noted in a handful of studies in the literature, there is no study to date that specifically focuses on examining people’s perception of regional variation in Cyprus. This study aims to fill this gap in the literature by studying the perceptions of thirty Greek Cypriots that live in the capital of Cyprus, Nicosia. In particular, it aims to answer to following research questions:

This study is part of a larger project which examines the linguistic performance and attitudes of Greek Cypriots who live in Nicosia. The main tool used in this project is the sociolinguistic interview in the original Labovian sense. Since we understand that “there is no such thing as unobserved language data” (Schilling 2013, 128), we have taken a number of steps in an effort to overcome what Labov (1972, 61) calls the ‘Observer’s Paradox’ and ensure that our participants did not aim to use a standard form (i.e., SMG) during their interviews but used the vernacular instead as much as possible. That is why, apart from trying to engage them in narratives of personal experience (Labov 1972), we have also decided to recruit only participants the interviewer (first author) was familiar with.

In an earlier study conducted in this setting, familiarity with the interviewer was one of the variables that affected how participants spoke during their interview. More specifically, it was one of the factors that contributed to generating more standard-like forms in their interviews (Tsiplakou, Armostis, and Evripidou 2016). This is something we wanted to avoid in our study. The recruitment of participants was conducted by adopting a judgement sampling technique. A judgement sampling technique involves recruiting participants according to a predefined set of criteria—in this case their age, sex and familiarity with the interviewer. It should be noted here that a judgement sample is the preferred type of sample in most sociolinguistic studies (Wardhaugh and Fuller 2015) and the minimum number of participants in each category created for the purposes of a study is five.

For our study, we needed to recruit 30 people in total since we created three different age groups and had to recruit 10 people for each age category, five females and five males. Our sample is balanced in terms of participants’ sex and age (Table 1). The participants were all residents of the city of Nicosia and our study can be replicated in the future with participants from other cities in Cyprus so as to compare and contrast the perceptions of people from different cities in this island.

As part of the language module of the sociolinguistic interview, our participants were presented with the map in Figure 4, which was designed by the authors for the purposes of this study. Following Garrett et al. (2003, 93), the participants were told to draw lines on the map in order to demarcate the geographical areas that, according to their own opinion and experience, depict the main regions within Cyprus were different dialects/ways of speaking are present. In other words, they were told to complete the so-called ‘draw-a-map task’ which, as explained in the introduction above, is one of the main tools through which people’s perceptions of spatial variation are examined within the field of perceptual dialectology (see e.g., Preston, 2013, and sources cited in the introduction of this paper). Our participants were also asked to label each area they drew, that is, give it a name of their choice. During the first interviews we conducted, we realized that this was a confusing question because there are no names ascribed for different regional dialects in Cyprus like there are in other countries, such as Germany or the United Kingdom, for example. Since the question confused our participants, we eventually stopped asking it. ¹ Subsequently, they were asked to justify their answers by naming the linguistic features they know of that each area uses differently. Finally, they were asked to name the area whose dialect has a wide social acceptability over the others, according to their opinion.

The map has no other information on it apart from black dots which stand for the main cities of the island of Cyprus corresponding to the six districts. It thus marks the districts of Cyprus without providing their names. As Preston points out, the kind of information researchers note on the map, such as state or district lines, major rivers and mountains, affects the kind of answers participants give Preston (1989, 25). That is why we decided to provide as little information as possible on the maps; hence we refrained from drawing the lines delimiting the districts and we did not provide their names either. One may argue that even the existence of the black dots could have affected our participants’ responses. We believe that not to be the case. Our experience with the participants during the interview was that many of them would comment on the places that they believe depict the presence of regional dialects before being shown the map—so the map did not really affect their responses. On top of that, many were relieved when they saw that the map had some marking on it because they found it easier to complete the task. Finally, as will become evident later in this paper, some participants marked areas on the map that had nothing to do with the marked cities on the map.

Our argument that it was better to provide as little information as possible on the maps finds support in the literature. In a similar study in the United Kingdom, Montgomery (2007) reports that results from perceptual dialect maps incorporating city location dots compared to those that did not showed no difference in the rate of frequency of a dialect area being indicated. That is, respondents marked the same regions on the map regardless of the type of map they were presented with (see Braber, 2015 for exactly the same observation). On top of that, Montgomery (2007) also notes that when presented with the maps with no dots, many of his participants complained that it made the task unnecessarily difficult (see also Preston 1989, 25). Further, in Evans et al. (2020, 3), regions that did not have cities marked on the map were marked by a high proportion of respondents, which suggests that their responses were not influenced by the markings on the map.

As noted above, the ‘draw-a-map’ task was part of the language module of a sociolinguistic interview. Before the participants were presented with the task, they had already responded to a series of questions on language, see Supplementary Material for the list of questions the participants were asked. One of these concerned regional variation in Cyprus and usually generated interesting discussions. The question was more often than not phrased in the form of “Can you understand that someone is from Nicosia just from the way they speak?” and sometimes in the form of “Can you understand from the way they speak where someone comes from?”. In response to the question, respondents would describe the areas where they believe people speak differently from them. Interestingly, when they would then be presented with the map, not all of them marked the same regions they had previously commented upon when answering the question. In fact, in an extreme case, one participant pointed out areas where people talk differently during the interview, but then refused to do the task claiming that the differences were not salient enough to be illustrated as different regional idioms on the map.

The fact that the ‘draw-a-map’ task was part of a recorded interview provided us with the opportunity to analyze both the results of the task as well as the participants’ metalinguistic comments made before and during the task, not just those they had written down on the map. The data generated by the task as well as the interview were analyzed in a variety of ways, both quantitatively and qualitatively, in order to answer the research questions. To answer RQ1 we calculated the areas drawn on each map and the recognition rate for each area following a methodology described in Montgomery (2016, 163). We first counted the lines drawn representing perceptual dialect areas on each respondent’s map. The percentage recognition rate was calculated by dividing the number of the lines drawn by 30 (the total number of participants) and then multiplying the result by 100 (see Dialect Drawn Areas). Then we examined and categorized the characteristics and evaluations of the different dialect areas offered by our participants following a coding system used in Long (1999) and Montgomery (2007) so as to address RQ2 (see Dialect Areas Characteristics and Evaluations).

The following procedure was followed in order to examine the characteristics and evaluations of the different dialect areas offered by our participants so as to address the second research question of the study. Upon examining the transcripts, we noted what the participants said about each region they mentioned both during the map task and prior, when they were asked to talk about regional variation in Cyprus in the course of the interview. In other words, we examined the comments the participants made both before and during the map task, irrespective of which areas they eventually decided to mark on the map. For example, participant 1 marked Nicosia, Larnaca and Limassol as one area on the map, but during their interview they also noted some characteristics they think are unique to Limassol. Following Long (1999) and Montgomery (2007), we categorized the characteristics and evaluations of the dialect areas that the participants provided in the following way:

In this section we answer the research questions we put forward in section 1 which we repeat here:

In response to RQ1, our participants identified 25 different areas with an average of four areas noted by each participant. Paphos is the area with the strongest geolinguistic salience with a recognition rate of 83%. This is not surprising given that comments on how people from Paphos speak differently from the rest of Cyprus are very frequent amongst Cypriots (Tsiplakou et al., 2006). The identification of Paphos as a separate regional dialect area and its strong geolinguistic salience is not only the result of certain (para)linguistic features that are characteristic of the area; it has a lot to do with stereotypical perceptions of this area as one that is remote, isolated, and away from the center. ⁵ Other areas that have a strong geolinguistic salience—but not as strong as that of Paphos—are the Ammochostos/Kokkinochoria area and Limassol with a 57% and 50% recognition rate, respectively.

The three areas that have a strong geolinguistic salience are also the ones that have been discussed the most when it comes to the linguistic and non-linguistic features that characterize their regional dialect. To answer RQ2, it is clear from Tables 3 and 4 that our participants mentioned more linguistic features to characterize these areas than non-linguistic ones. Overall, it is believed that the regional dialect of Paphos is mainly characterized by prosodic differences, the area of Ammochostos/Kokkinochoria by phonetic ones, with a clear emphasis on two variants [ç] and [c], and Limassol by lexical differences. Most of the linguistic features mentioned for these areas are indeed used at least by some speakers that live there. A notable exception is the claim that people in the Ammochostos/Kokkinochoria area use [c] instead of [tʃ] before [i] and [e] when using CG. This is, in other words, a false stereotype. Interestingly, some participants claim that they have even heard people in those areas using [c] before [i] and [e] instead of [tʃ] which clearly shows that “stereotypes may be strong enough to overcome linguistic evidence” (Preston 1993, 182). The claim that [c] is used in that community’s lect is a great example of how the folk make sense of linguistic variation that does not make sense to them otherwise. As explained before in this paper, the confusion originates from the use of [ç] instead of [ʃ] before [i] and [e] in that community which is understood as an effort by those speakers to use SMG and not as variation within CG. The false claim that these speakers also use [c]—another SMG variant—instead of [tʃ] is a way to make sense of this “paradoxical” situation: it reinforces that claim that these speakers somehow aim to/want to use SMG but fail to do so. In other words, for this narrative to make sense, the use of [c] is added to the perceptual linguistic repertoire of these speakers.

In response to RQ2, our participants refrained from making evaluative comments and labelling the regional varieties and their speakers negatively, with a few exceptions (see Table 4). In Evans et al.’s (2020)f study where this phenomenon was also observed—contra many studies in the literature some of which were set in the same setting as that of Evans et al.’s (2020) study—the authors acknowledge that “more evaluative labels […] were expected” (p. 5) and conclude that “further research is needed to understand this finding” (p. 6). As it has been acknowledged in this paper, this finding may be partly the outcome of the kinds of questions our participants were asked and of the effect of the social desirability bias (see also Evans et al., 2020). The lack of many evaluative comments may also have to do with the fact that for many of the participants the differences noted for each dialect area were not many or salient for them. One exception is the phonetic differences noted for the Ammochostos/Kokkinochoria area which when evaluated where mostly evaluated in a negative light (as strange and funny). The lexical differences noted for Limassol, on the other hand, were noted as mere differences; they were not viewed in a negative light. More studies are needed in this context in order to further understand this finding and make fruitful comparisons.

RQ3 was a question that our participants were asked directly towards the end of their interview. The majority of them stated that they consider the regional dialect of the central part of Cyprus to be more widely accepted than that of the periphery, with the variety of Nicosia, and sometimes the areas or cities next to it, considered by many to be a more prestigious and widely accepted form of speech. Interestingly, only one third of the participants think that their own city constitutes a distinct dialect area and most participants consider their way of speaking to be neutral without certain linguistic features that characterize it (see Table 3). The two findings are not unrelated since the perceived absence of marked features can be seen as an indication that the variety of the central area of Cyprus is considered to be the standard and accepted one (Milroy 2001), also see Neises (2013) for a similar finding.

In response to RQ4, i.e., whether speakers are aware of dialect levelling processes in their community, the results of this study indicate that many of the participants are consciously aware of what Kerswill (2003) and Torgersen and Paul (2004) call regional dialect levelling, which refers to the “the levelling of differences among what was at first a conglomeration of varieties, often leading to a new variety characterized by the absence of localized forms” as the “outcome of various partly geographically-based language change processes” (Kerswill 2003, 224). ⁶ These are the participants who refused to draw the map because they do not think there are that many differences nowadays in how people from different areas speak or they do not regard such differences to be salient and the participants who drew a map of the past because this is when they believe differences existed. Many participants also showed how well aware they are of regional levelling in Cyprus through their metalinguistic commentary, examples of which have already been provided (see 20 and 21). More examples follow below.

The above examples show that these participants not only know that regional dialect differences have diminished in today’s CG, but they are also aware of some of the reasons why. Finally, the fact that participants drew an average of just four areas on their maps can also be seen as evidence of their awareness that there is no longer dense regional variation on the island.

It is worth nothing that a few participants also demonstrated that they are also aware of the process of levelling as defined in the narrow sense by Kerswill (2003) and Torgersen and Paul (2004), see footnote 8. The two examples below are particularly interesting.

Participant 16 talks about how others (here people from Limassol and the area of Kokkinochoria) need to change and accommodate their way of speaking when relocating to Nicosia because through their interactions with other speakers the local variants they use become salient and marked. Participant 6 explains what it feels like when one uses a local variant that is a cause of ridicule by other speakers: one eventually needs to accommodate to how the majority speaks if they do not want other people to comment on and make fun of how they speak.

The final question to address in this paper (RQ5) concerns the implications of the findings of this study for research on language variation and change. To start with, some of the findings of and observations made in this study give support to the claim that CG is undergoing levelling of marked local basilect features and subvarieties. This shows how studies on perception can be used along study on production to study the same phenomenon and inform one another. This is, for example, shown by the fact that participants who completed the ‘draw-a-map’ task tended to view the different cities and districts, or combinations thereof, as the areas where there is regional variation. The emphasis on urban sites is in line with many recent studies in the literature (Al-Rojaie, 2020; Evans, 2013b; Evans et al., 2020; Montgomery, 2007; Montgomery, 2016). Montgomery (2007, 2016) who also found that the identification of city-based names for dialect regions to be frequent in his study of perceptions of English in the United Kingdom, notes that this is a relatively new phenomenon in British perceptual dialect geography and suggests that this city-based dialect awareness is consistent with regional dialect levelling. Moreover, the participants’ linguistic description of regional variation lacks detail; it is superficial [see Bounds (2015) and Burbano-Elizondo (2008) for similar observations]. The fact that many speakers could not clearly describe the features of the perceived subvarieties they reported, could not give examples, or gave only a couple is, as Themistocleous and Tsiplakou (2013, 448) argue, a “very good indirect evidence for levelling” (see also Tsiplakou et al., 2006).

Studies on language perception can inform work on language production and contribute to the research of language variation and change in fruitful ways (Preston 1993; Gordon 2013; Meyerhoff 2016; Cramer 2021). This study has shown that many participants are aware of regional dialect levelling in their community, and a few are even aware of the more specific process of levelling and accommodation as the mechanism behind it. If it is indeed true that due to the invasion, “abrupt levelling must have taken place within the space of one generation” (Tsiplakou et al., 2016, 11), then it is not surprising that people born before 1974 are aware of regional dialect levelling. Yet, interestingly this study has also shown that some younger speakers are also aware of regional dialect levelling in Cyprus. Other studies in the literature do not really discuss speakers’ awareness of regional dialect levelling in a given community. Perhaps future studies can examine this aspect of linguistic perception in order to understand under which circumstances speakers are aware that regional dialect levelling is underway in a community. For example, it might be possible that the size of the community and the size of the area under study plays a role. Future studies of small communities and small geographical areas within a perceptual dialectology framework should be conducted in order to examine the role that the size of the community and place plays.

It has also been shown that, while the participants’ linguistic description of the dialect areas they perceive to exist in Cyprus was not rich or supported with many different examples, it was mainly accurate. In other words, people’s perceptions of linguistic variation in Cyprus seem to be generally correct despite the presence of a few false stereotypes surfacing in some speakers’ metalinguistic commentary. This study has also shown that research on perception can inform claims made by language studies on production, since some of the findings of the present study can give credit to the claim that CG is undergoing levelling of marked local basilect features and subvarieties. Of course, given the small sample studied here, further research with more speakers from other areas of Cyprus is needed to confirm these findings.

Overall, it can be argued that this study has shown the need to conduct more research on language perception in this community. As it has been noted at the beginning of this paper, a CG koine is developing in this community. This variety is based on the variety of the Mesaoria region—the island’s central plain—and it has been argued that it enjoys prestige amongst Greek Cypriots. People in Nicosia seem to ascribe some type of prestige to their language and generally to the language of the central parts of Cyprus. It is unclear whether they do so because they just value their own way of speaking or because their variety is more or less what the CG koine is based on. Studying whether people from other regions in Cyprus share the view that the language of the central parts of Cyprus is more widely accepted and prestigious than that of the periphery will help us answer this question. If it is shown that people in other areas tend to find their own way of speaking more accepted than others’, this will be an indication of people’s value of local speech and identity which may act as resistance to the processes behind regional dialect levelling.

Finally, it has been argued in the literature that features of older local basilects that are still used post-levelling in the community appear to operate as indexicals of lower registers (Papapavlou and Sophocleous, 2009; Tsiplakou et al., 2006, cited in Tsiplakou and Armostis, 2020; see also discussion of example 1 in Tsiplakou and Armostis, 2020, 373). While this is indeed true, it is clear from some of the data in the present study that geographical space is also important in people’s perception of linguistic variation on the island. More specifically, some (basilectal) features clearly index specific regional varieties and specific places in Cyprus [e.g., (ç) before (i) and (e) as a stereotypical characteristic of the Kokkinochoria region and a special intonation contour as a stereotypical characteristic of Paphos], while the use of basilectal features can indicate that someone comes from a rural area, as shown in example 23. It is clear that more research on language perception is needed in this community to fully understand the current sociolinguistic situation and to uncover the social meanings and indexical values of certain linguistic features.