Representing Variation in Creole Continua: A Folk Linguistic View of Language Variation in Trinidad

1. Introduction

Variation in creole continua has traditionally been described from a structuralist perspective concerned with establishing underlying grammatical systems. As a result, creole studies have tended to reduce variation in creole continua to a structural distinction between acrolectal and basilectal varieties, in between which a typologically intermediate mesolectal variety is sometimes acknowledged. Relying exclusively on this taxonomy runs the risk of imposing categories that are not acknowledged within the relevant speech community; at the same time, it possibly ignores varieties not necessarily signaled by distinctive grammatical features. Additionally, it produces a unidimensional picture of variation in which varieties are ranked only in terms of their degree of standardness, while indications exist that variation in creole continua can be simultaneously observed along other dimensions. This article takes the view that, in order to ultimately arrive at a representative picture of variation in creole continua, account must be taken of not only production data, but also of locally perceived linguistic categories. Following a protocol developed within Perceptual Dialectology (see, e.g., Preston 1999), this article proposes to describe perceived linguistic distinctions via the elicitation of spatially contextualized folk linguistic descriptors. The case study that it relies on involves the Caribbean island of Trinidad, where Standard English historically co-exists as an official language with creolized varieties of English that creolistic accounts refer to as “Trinidad English Creole” (see, e.g., Mühleisen 2013). The specific question that this article is raising is twofold: what are the linguistic categories perceived by non-linguist Trinidadians, and what are these linguistic categories likely to tell us about social and sociolinguistic dynamics in Trinidadian society?

This article is organized as follows. Section 2 summarizes views on objective patterns of variation in creole continua and reflects on the extent to which these patterns are perceived. Section 3 presents Trinidad’s sociolinguistic ecology and local sociolinguistic data. Section 4 describes the data collection protocol, followed by a general breakdown of the data and the methodology that the study uses for analyzing these data. The analysis is conducted in two stages. Sections 5 through 9 focus on the descriptors formulated by the respondents. The second stage (section 10) attempts to match the data with existing creolistic accounts, while qualifying their possible correlations with actually observable variation patterns.

2. Collecting Perception Data on Variation in Creole Continua

Originally introduced by DeCamp (1971) and Bailey (1966), and since then highly visible in creole linguistics, the notion of creole continuum ¹ generally refers to the space between restructured varieties and their lexifier languages. ² Of how many varieties a creole continuum is comprised has proven a thorny issue. DeCamp (1971) describes variation in creole continua in structural terms of implicational rules operating along a “seamless cline” between one creole (or “basilectal”) and one superstratal (or “acrolectal”) pole, of which individuals or communities may command different spans (see further DeCamp 1961). ³ Other structuralist perspectives that have come to the fore have been concerned with establishing clear-cut distinctions within creole continua by categorizing linguistic features as acrolectal or basilectal, and in the case of typologically intermediate features, as mesolectal (Bickerton 1973; Winford 1997). The rural-urban divide has been consistently used as a socio-spatial yardstick for distinguishing between putative basilectal features, which have typically been sought in rural areas, and mesolectal or acrolectal features, which have typically been associated with urban areas following the rationale that urban areas are more likely to show convergence with established normative models (Rickford 1987; Romaine 1988; Hellinger 1998). Even though they have been tested in quantitative studies of language variation (cf., Winford 1972; Patrick 1999), structural taxonomies have been criticized for reducing variation to structural distinctions along a unidimensional scale ranging from more to less standard while potentially overlooking other linguistic distinctions—not necessarily structural nor measurable in terms of standardness—that are actually perceived in creole-speaking societies (cf., LePage 1980; Carrington 1993; Irvine 1994). In line with this criticism, this article generally argues that emic perceptions of language variation can fruitfully be used as one possible point of departure in the search for linguistic distinctions within creole continua.

A glance at the scattered comments that early creolistic studies made on sociolinguistic distinctions within creole continua suggests that it is unfair to deny any social relevance to the structural distinction between acrolect and basilect. Not identifying with the watertight character of this distinction, DeCamp (1971:341) himself observed in his time that “many Jamaicans persist in the myth of two varieties, the patois and the standard.” Rickford (1987:22) similarly noted a perception among his Guyanese informants that Guyana’s linguistic landscape was marked by a single distinction between “English” and “Creolese.” Yet, there is evidence that variation in creole continua can also display a multidimensional character, whereby alternating linguistic forms may rank equally in terms of their [+standard] or [+creole] features. Such variation has been documented across, among other things, alternating footings (e.g., Washabaugh 1977; LePage 1980), space (e.g., DeCamp 1961; Singler 1987), social statuses (e.g., Irvine 1983; Singler 1987), and ethnicities (e.g., Escure 1982; Irvine 1994). In all cases, the observed patterns of variation are perceptually associated with specific social identities or personae, or in other words, they are endowed with social indexicality (cf., Silverstein 2003; Johnstone, Andrus & Danielson 2006). Among the studies that most systematically sought to establish social indexicalities, Irvine (1994) administered a matched-guise test to Jamaican respondents with a view to confirm perceptual associations between specific phonological patterns in Jamaican English and white Jamaicans. While Irvine (1994) relied on her intuitions for conceptualizing this social category and for assuming that its sociolinguistic saliency can be established, this study proposes to identify the sociolinguistic stereotyping processes likely to warrant such intuitions, treating them as possible indicators of where salient linguistic distinctions can be located in a given creole continuum.

Sociolinguistics has established that social distinctions often coincide with linguistic distinctions. Their correlations are conventionally encapsulated in a finite range of independent variables: geographic location, age, social class, gender, ethnicity, networks, and communities of practice feature most notably among these variables (Labov 2001; Meyerhoff & Strycharz 2013). As a result, some group-specific linguistic features come to develop social indexicality, which varies in terms of its saliency (cf., Silverstein 2003). The most salient indexical relationships essentially form the outcome of a stereotyping process whose rationale is closely intertwined with the imperative of operating a distinction between “us” and “them” (Labov 1972; Schneider 2004). For that distinction to be operationalized, differences between ingroups and outgroups are typically emphasized, sometimes reified into symbolic boundaries expressed by opposable linguistic labels, often endowed with connotations of spatial boundedness and internal linguistic homogeneity, whose emergence and currency is determined by sociolinguistic metadiscourses on identity (cf., Garner 2004; Johnstone, Andrus & Danielson 2006; Agha 2007; see further Barth 1969). Perceptual Dialectology (hereafter “PD,” see Preston 1999; Long & Preston 2002) offers a methodological framework for eliciting perceived sociolinguistic distinctions. The fact that it characteristically concerns itself with spatialized sociolinguistic perceptions makes it particularly suitable for a study of creole-speaking societies, where the notion of an urban-rural divide seems to underlie a strongly entrenched stereotypical perception of how standard and creole varieties are socially distributed (cf., Rickford 1987; Romaine 1988). This article proposes to use one specific PD protocol, namely, the map-drawing and labeling task, to establish how generalized unidimensional representations of variation are while also offering scope for eliciting perceived distinctions within standard and creole varieties.

3. Socio-Spatial and Linguistic Distinctions in Trinidad

Formerly under Spanish rule (1592–1797) before becoming a British colony (1797–1962), Trinidad is an ethnoracially diverse Caribbean island whose population of little more than a million inhabitants is polarized between an Afro-descendent component (34.2 percent of the total population) and an Indo-descendent component (35.4 percent). Individuals claiming a mixed ethnic ancestry make up 22.8 percent of the total population, among whom one third claim a mixed Indo- and Afro-Trinidadian ancestry. The remaining census categories all stand below 1 percent (CSO 2012). Most of Trinidad’s population is concentrated in the Northwest, while the rest is mainly concentrated in the cities of San Fernando and Chaguanas (CSO 2012). There is a spatial dimension to ethnicity in Trinidad visible in regional patterns of ethnic concentration (Goodenough 1978; Clarke 1993). The central and southern municipalities are home to populations that are predominantly of East Indian descent. Together, they make up the historical “Sugar Belt” where East Indian laborers were settled by the British in the nineteenth century (Brereton 1974; Singh 1974). By contrast, the Afro-descendent population—who can be traced back to the Martinican slaves imported by French planters invited by the Spanish Crown and to subsequent immigration from the British West Indies under British rule—forms the largest population group in the northwestern region. Populations of mixed ethnic background are most strongly represented along the transport corridor that spans the Northwest and Northeast, forming the majority in the latter area (CSO 2012; see further Brereton 1979). An overview is given in Figure 1. By the oil boom in the 1970s a considerable gap was still observable between Trinidad’s small white population—who earned the highest incomes—and others, as well as between Afro-Trinidadians and Indo-Trinidadians, the latter then still forming the poorest segment of the population (Henry 1993). Socio-economic and ethnic differentials are simultaneously reflected in the urban landscape. Some economically disadvantaged urban areas, such as Laventille and Morvant (see Figure 6), are perceptually associated with Afro-Trinidadians; Port of Spain’s wealthy western suburbs, on the other hand, are associated with Trinidadian whites and expatriates (Ramdhan 2010; Prentice 2015).

OPEN IN VIEWER

Figure 1.

Local Proportions of Ethnic Categories in Trinidad

Trinidad’s various population groups have by now overwhelmingly shifted to Trinidad English Creole (TEC), which has developed in a sociolinguistic context dominated by Standard English (SE). Patterns of language variation in Trinidad have mostly been discussed in terms of grammatical systems. Based on quantitative data, Winford (1972; 1997) argues that only two linguistic systems co-exist in Trinidad, namely SE and TEC, which operate in discrete social domains, defined by context, social class, location, age, and ethnicity: [+basilectal] features correlate more with informal contexts, lower working classes, rural areas, older age cohorts, and Indo-Trinidadian ethnicity; [-basilectal] features correlate with formal contexts, middle/upper working classes, urban areas, younger age cohorts, and Afro-Trinidadian ethnicity (see further Solomon 1993:183-184). Winford (1997) emphasizes that TEC is a relatively homogeneous grammatical system, and that its interaction with SE results in various degrees of code-switching that show no signs of giving rise to an intermediate mesolectal system. Support for Winford’s (1997) claim could be found in the fact that Trinidadians tend to perceive a distinction between SE and what they increasingly tend to refer to as “Creole” (Mühleisen 2001). However, account must be taken of the possibility that grammatical systems do not necessarily coincide with perceived varieties. As a result, there is a possibility that the descriptor “creole” may also generically refer to various modes of combining SE and TEC, described in detail by Youssef (1993; 1996). Rather than as combinations of two perceptually bounded varieties, these code-switching/mixing patterns may be perceived as distinct varieties of their own in the same way as specific patterns of code-switching elsewhere have been observed to be (see, e.g., Chen 2015; Borges 2017).

One possible point of departure in the search for linguistic boundaries in Trinidad is to administer map-drawing and labeling tasks as suggested in section 1. What could be gained from them is indications of whether Trinidadians perceive a sociolinguistic distinction between urban and rural areas whereby SE is associated with the former and “Creole” with the latter in keeping with the assumptions traditionally made in the creolistic literature (see section 2). Additional insights into perceived sociolinguistic boundaries in Trinidad could come in the form of elicited perceptions of sociolinguistic distinctions within urban and rural areas. I am presenting in the following section the map-drawing and labeling task devised for the purpose of this study and the sample of informants that it was administered to. Additionally, I present the methodology I employ for analyzing the perceived sociolinguistic distinctions that these tasks produced.

4. Data Collection and Analysis

4.1. Data Collection

The origin of Perceptual Dialectology is closely intertwined with the use of perceptual maps of dialect isoglosses (cf., Preston 1999; Rensink 1999). By additionally prompting respondents to label or describe the isoglosses they draw, access is potentially gained to local linguistic taxonomies as well as to stereotyped representations of linguistic usage, as illustrated by, among other things, the US dialect perception data discussed by Preston (2003) and Bucholtz et al. (2007). Non-linguists tend to depict language varieties in terms of proximity to or distance from their own linguistic variety (see, e.g., Grootaers 1999; Benson 2003). As a result, obtaining a picture of sociolinguistic perceptions that can be called representative of a relatively wide geographic area calls for a demographically diversified sample of respondents. Such an approach is likely to produce different perspectives that can reduce the effects of own variety bias, as illustrated by, among other things, Long’s (1999) study of subjective Japanese dialect divisions. Once the respondents’ perceptions are elicited, isoglosses and descriptors can be collated in order to arrive at generalizations. While it is tempting to search perception data for suggestions of objectively verifiable sociolinguistic boundaries, it is important to keep in mind that sociolinguistic perception data relies to a large extent on stereotyping, and may or may not reflect linguistic reality as a result (Hundt, Anders & Lasch 2010; see further Grootaers 1999). Rather, perception data are likely to offer glimpses of local ideologies of sociolinguistic stratification, which in the context of map-drawing and labeling tasks will have a spatialized dimension (see section 1). In more concrete terms, spatialized perception data in the Trinidadian context are likely to reveal where prestige norms are located, and what geographic, social, and linguistic attributes partake in sociolinguistic stereotyping processes in the context of Trinidad.

In keeping with the protocol used by Bucholtz et al. (2007), the research instrument used for this study consists of a photocopied outline map of Trinidad with general survey questions. The top of the outline map features the instruction: “Please draw a boundary around each part of Trinidad where you believe people speak differently, and label/describe the area as you see fit.” General survey questions concerned gender, age, ethnicity, and place of birth/longest residence. The sample of respondents consists of University of the West Indies students aged between 19 and 24. A total of one hundred maps were collected, of which three examples are provided in Figures 2, 3, and 4. The sample of selected respondents is intended to reflect Trinidad’s population as much as possible in terms of gender and ethnicity. Students from Tobago and foreign-born students were excluded. Additionally, the selection excluded students whose curriculum involved subjects related to linguistics or education sciences in order to limit the chances of familiarity with linguistic terminology that could bias the data. The sample comprises both males and females (54 percent and 56 percent, respectively). In terms of ethnicity, 38 percent of respondents are Afro-Trinidadian, 26 percent Indo-Trinidadian, 34 percent “mixed,” and 2 percent “white.” Based on the places of birth/longest residence they declared, the respondents are categorized as either “Northern” (38 percent) or “Southern” (62 percent), whereby “Northern” refers to the northernmost two administrative divisions shown on Figure 1, and “Southern” to the rest. The rationale behind stratifying the sample into Northern and Southern categories is that the demographic contrasts between North and South are expected to underpin generic “us-them distinctions” (section 3).

OPEN IN VIEWER

Figure 2.

Sociolinguistic Boundaries in Trinidad According to a 22-Year-Old Southern Indo-Trinidadian Male Student

OPEN IN VIEWER

Figure 3.

Sociolinguistic Boundaries in Trinidad According to a 21-Year-Old Southern Female Student of Mixed Ethnicity

OPEN IN VIEWER

Figure 4.

Sociolinguistic Boundaries in Trinidad According to a 19-Year-Old Southern Afro-Trinidadian Female

The example maps (Figures 2 and 3) illustrate two tendencies among the respondents. One is to draw boundaries between perceived linguistic areas, while the other is to point out specific locations perceived as linguistically salient (Figure 4). The example maps give a first impression of the range of descriptors used by the respondents. While broad linguistic labels are sometimes used, such as “Creole” or “English,” mental representations of distinctive varieties sometimes seem implicitly subsumed under geographic or social descriptors. Examples are found on Figure 2 where the general social descriptor “high society” and the attitudinal descriptor “proud” are applied to the island’s northwestern corner (which the respondent delimits without assigning a geographic descriptor to it). Social descriptors also tend to involve an ethnoracial specification, such as “Bush Indian,” but ethnoracial descriptors can also be used autonomously, such as “White” or “Indian” (Figure 4). The respondents’ comments occasionally convey holistic descriptions of individual repertoires. On Figure 3, the respondent associates the ability to keep “Standard English” and what she calls “Broken English” apart in communication with a region she labels as “East” (see further sections 5 and 9). In other cases, the respondents do not dwell on modes of cohabitation between varieties. Figure 2, for example, applies the descriptors “Creole” and “broken” to Trinidad’s northeastern corner, possibly reflecting a perception that no varieties other than creole/broken are used by or known to local residents of that area (see further section 9).

4.2. Data Analysis

Before embarking on the process of data analysis, parameters have to be defined for matching isoglosses and descriptors produced by the respondents. In keeping with Bucholtz et al. (2007), each isogloss and descriptor is fit into predefined spatial units that are referred to as “subregions.” Each of those subregions, shown in Figure 5, consists of combined census tracts in order to facilitate the social contextualization of the descriptors produced by the respondents. While combining census tracts social boundaries suggested by the socio-historical data described in section 3 are as much as possible respected. As shown in Figure 5, the list of subregions is as follows:

OPEN IN VIEWER

Figure 5.

Census Tract Outlines Used for Contextualizing Respondents’ Descriptors

I. “Northwest”: The northwestern tip of Trinidad comprising Port of Spain, the capital (census tracts 1-2).

II. “Midnorth”: Census tracts 3-4, home to an Afro-Trinidadian majority, which cover, among other things, the East-West corridor, a sprawling urbanized area along Trinidad’s East-West highway that constitutes the island’s largest conurbation.

III. “Northeast”: Census tract 5, which is less urbanized than its western neighbor, and home to an ethnically “mixed” majority.

IV. “Central”: Census tracts 6-7, which cover a large semi-urban area centered on the centrally located mid-range city of Chaguanas, home to an Indo-Trinidadian majority.

V. “South”: Census tracts 8-13, which cover a region home to an Indo-Trinidadian majority, as well as to San Fernando, the island’s second largest city (census tract 10).

Descriptors clearly related to dots placed within a specific census tracts were counted as linguistic attributes of the same subregion comprising the same census tract. Descriptors which are unrelated to dots and instead fit into linguistic areas straddling two or more subregions were entered into the database as attributes of each of the relevant subregions. A quantitative analysis was conducted for the purpose of establishing which of the five subregions were most frequently labeled or described by the respondents. Next, the descriptors provided by the respondents were ordered into generic categories comparable to those that Bucholtz et al. (2007) used for their Californian PD study (see Table 1). The most frequent category consists of spatial descriptors (e.g., “South,” “San Fernando”). The second most frequent category consists of linguistic descriptors (e.g., “proper English,” “yankee talk”). In third position come social descriptors (e.g., “bush Indians,” “ghetto”), followed by specific linguistic features named by the respondents as characteristic of local linguistic usage (e.g., “my one” for the possessive pronoun mine, “what up” for the expression what’s up).

OPEN IN VIEWER

Table 1.

Types of Descriptors

Spatial	516 (41.6%)
Linguistic	505 (40.8%)
Social	155 (12.52%)
Specific examples of linguistic usage	14 (1.13%)
Other	6 (0.48%)
Total	1238 (100%)

Note: I include in the term descriptor multiword characterizations that read as sentences.

The analysis attempts to identify the perceived components of the Trinidadian creole continuum based on levels of consistency across perceived sociolinguistic distinctions. While the linguistic significance of descriptors is interpreted, specific attention is paid to holistic characterizations of perceived patterns of linguistic diversity at a local level, including stylistic repertoires and their conversational manifestation. The final part of the analysis attempts to match the obtained perception data with production data elicited in the Trinidadian context.

In the next sections I first discuss spatial and social descriptors (sections 5 and 6). Sections 7 to 9 focus on linguistic descriptors. In these sections attention is paid to the generic distinction that respondents make between standard and non-standard varieties (section 7), to perceptions of linguistic differentiation within the standard and non-standard categories (section 8), and to linguistic characterizations that refer to modes of cohabitation between distinct varieties (section 9). Finally, section 10 collates all descriptors and situates them in the context of production data elicited by other studies and of views on variation in creole continua presented in section 2.

When a specific respondent has to be quoted, that respondent is referred to by means of identifying numbers adjoined to prefixes indicating their respective self-declared ethnoracial background (“IT” for Indo-Trinidadian, “AT” for Afro-Trinidadian, and “MI” for mixed ethnoracial background), their geographic origin (“N” for North and “S” for South), and their gender (“M” for male and “F” for female). For example “IT-S-F1” refers to a female Indo-Trinidadian respondent from the South, and “AT-N-M1” refers to a male Afro-Trinidadian respondent from the North.

5. Spatial Descriptors: North and South, Cities and “Bush”

The maps tend to relate linguistic information to drawn areas that may or may not be assigned spatial descriptors (section 4.1). Aside from delimiting areas, a widespread respondent strategy for spatializing information relies on only naming specific urban locations (i.e., “Chaguanas,” “San Fernando,” etc.). Both strategies combine into signaling generic linguistic distinctions perceived by the respondents. The most salient spatial distinction is one that involves a North/South divide, which is signaled by 93 percent of the one hundred respondents by means of drawn boundaries and/or by the simultaneous naming of northern and southern urban locations. The next most salient spatial distinction, made in 68 percent of cases, is one that marks out the central subregion. Finally, 62 percent of respondents signal an East/West divide. The southern respondents perceive an East/West divide less frequently than do the northern respondents (53 percent versus 76 percent). Similarly, the southern respondents are less inclined to perceive a central region than their northern peers (61 percent versus 76 percent). The descriptors used to refer to the drawn areas almost always include the name of a cardinal point. The geographic concept of an “East-West corridor” is used in only four cases (section 4.2).

The respondents tend to name several urban locations within the spatial units that they perceive. Table 2 displays the average number of urban locations named within each subregion, and shows the answers for southern and northern respondents separately. The most frequently named urban locations are Port of Spain, the capital, followed by San Fernando, the largest city in Trinidad’s southern part (Table 3; Figure 6). The figures suggest that the range of named urban locations correlates with the geographic origin of the respondents: northerners name more urban locations within the northwestern subregion than do their southern peers, whereas southerners name more urban locations within the southern subregion than do their northern peers.

OPEN IN VIEWER

Table 2.

Average Numbers of Urban Locations Named by Respondents within Subregions

Subregion	Average number of urban locations named	Average number of urban locations named by the Southern respondents	Average number of urban locations named by the Northern respondents
Northwest	1.11	1.06	1.28
Midnorth	0.36	0.28	0.52
Northeast	0.26	0.25	0.38
Central	0.27	0.25	0.28
South	1.74	2.39	0.28

OPEN IN VIEWER

Table 3.

Most Frequently Named Urban Locations

Subregion	Name	% of all locations named
Northwest	Port of Spain	9.9%
South	San Fernando	9.4%
Northeast	Toco	6.5%
South	Penal	5.9%
Northwest	West Moorings	5.6%
South	Debe	4.2%
South	Moruga	4.2%
Midnorth	Arima	3.9%
Central	Chaguanas	3.9%

OPEN IN VIEWER

Figure 6.

Most Frequently Named Urban Locations

The urban locations most often named by the Northern respondents are Port of Spain (13.79 percent of all urban locations that they named), followed by Laventille (10.34 percent), Arima (6.89 percent), Chaguanas (5.17 percent), and Blanchisseuse (3.44 percent). These locations are concentrated in the three northern subregions, except for Chaguanas, which is located in the central subregion. Laventille and Arima are municipalities connected to Port of Spain via the urbanized East-West Corridor, while Blanchisseuse is an isolated rural settlement on Trinidad’s northern coast. The locations most often named by the southern respondents are San Fernando (11.52 percent of all locations they named), followed by Port of Spain (7.81 percent), Penal (7.4 percent), Debe (5.76 percent), and Moruga (5.34 percent). Except for Port of Spain, these locations are concentrated in the Southern subregion. In contrast to San Fernando, Penal, Debe, and Moruga are small rural settlements. The contrasts in the ranges of urban locations named by the northern and southern respondents reflect the ingroup bias associated with stereotyping processes (Schneider 2004). This bias is reflected in the tendency for non-linguists to perceive more linguistic distinctions within familiar adjoining areas than farther afield (Preston 1999). Still, the respondents largely agree on the major linguistic divides that they perceive (i.e., North versus South, West versus East), as well as on the notion of a central region.

6. Social Descriptors: “Yankees” and “Bush Coolies”

A total of forty-eight maps provides social contextualization in the form of descriptors that refer to social groups or to stereotyped behaviors. In many cases, it is not clear whether these social descriptors are primarily meant to provide a social description, or a socially contextualized linguistic one. Some respondents use ethnoracial terms in the form of free-standing adjectives (e.g., Indian, White), which might refer to the ethnic character of the region, or to ethnic ways of speaking. The term “Indian” occurs on eleven maps (three of which come from Indo-Trinidadian informants, four from Afro-Trinidadian informants, and three from mixed informants), and is generally applied to southern areas in which Indo-Trinidadians form a demographic majority (section 3). While it may be used on its own, Indian may be added as a specification, such as in “bush Indians.” “Indian” seems interchangeable with the derogatory term “coolie,” used on its own on four maps (two of which come from Indo-Trinidadian informants). This latter descriptor is generally applied to rural areas in the southern and central subregions, and more specifically to the Indo-dominated settlements of Penal, Debe, Icacos, Mayaro, and Couva (section 5). Among the other ethnoracial descriptors features the label “White,” which appears on three maps as an attribute of the Northwest, more specifically of Port of Spain and of its affluent suburb of West Moorings. Interestingly, the respondents do not use any ethnoracial descriptor referring specifically to Trinidad’s Afro-Trinidadian population (see further section 8).

Other social descriptors are spatial descriptors with social connotations. In particular, the terms “bush,” and secondarily “country,” are used to refer to rural areas of Trinidad (“bush” occurs on seventeen maps, and “country” on four, in one case in combination with “bush”). Another descriptor with a socio-spatial meaning is “ghetto,” used twenty-seven times to refer mostly to the northwestern municipalities of Laventille and Morvant. Other, less frequently used, social descriptors are the terms “upper-class” and “middle-class” (used on two maps). The former is associated with the affluent northwestern suburbs, as well as some areas of San Fernando (i.e., Gulf View, Palmiste), and the latter is used once to refer to St. Augustine, a municipality in the midnorthern subregion. Finally, names of professions are occasionally used to characterize areas. For example, the northeastern subregion is associated with “farmers” in one case (EA-N-F2), the Cedros area with “fishermen” (AF-S-F6), and the Northwest with “business people” (MI-N-F4).

Another category of social descriptors involves references to social behaviors, attitudes, or lifestyles. The adjective “stush,” synonymous with posh, is applied in twelve cases to the Northwest, as well as to San Fernando, alongside miscellaneous adjectives that include “stuck up” and “bourgeois.” These adjectives have counterparts applied to the opposite end of Trinidad’s socio-economic hierarchy among which feature “uneducated” (four times) and “wild” (three times), applied more specifically to the northeastern and southern subregions, and “gangsta,” applied by EA-N-F2 to Laventille and Morvant. Some behavioral/attitudinal descriptors occasionally imply an accusation of illegitimacy. For MI-S-M3, residents of Chaguanas in the central subregion “feel better than everyone else,” while other respondents use the term “wannabe” in combination with various descriptors, such as “ghetto,” to refer to locations outside the Northwest. For MI-N-F7, residents of the Northwest are “fake,” and AF-S-F11 as “wannabe American” in the view of AF-S-F11. Other derogatory behavioral/attitudinal descriptors referring to social behaviors include “loud and aggressive” (MI-S-F3; describing the Northeastern subregion), “rough and tough” (MI-S-F2; describing Laventille), “crass,” and “crude” (AF-S-F13; describing Port of Spain).

To sum up, the respondents’ social descriptors fall into four distinct themes, namely, ethnicity, rurality, social class, and (in)authenticity. Social descriptors without immediate linguistic connotations are a recurrent feature of linguistic map-labeling tasks, as exemplified by, among other things, the descriptors “friendly” or “materialistic Southern Californians” reported by Bucholtz et al. (2007). The implicit imputation of linguistic saliency to social attributes manifests an “iconization” process associated with the development of social indexicality (Irvine & Gal 2000; Preston 2010). The next three sections discuss what explicit linguistic distinctions result from this iconization process, and what linguistic features are involved in it.

7. Polarized Linguistic Descriptors: The Northwest versus the Rest

Ninety-one maps display linguistic descriptors. The most widespread term featuring in these descriptors is the term “English” used as a noun (117 occurrences out of 505 linguistic descriptors, fifty-one respondents). The noun “English” is in all cases combined with a qualifier, which include, in order of frequency, “standard” (thirty-three occurrences), “broken” (thirty-five occurrences), “proper” (seventeen occurrences). Less frequently used qualifiers are “urban,” “rural,” and “Creole.” Twenty-one respondents out of the fifty-one who use the term “English” as a noun use it consistently to refer to all varieties between which they distinguish. They specify these varieties with the help of qualifiers: “Standard English” and “proper English” are opposed to “broken English” on the same maps. The descriptors “Standard English” and “proper English” exhibit a strong correlation with urban areas: in twenty-four out of fifty cases, they are related to the urban Northwest, and in eleven cases to San Fernando. ⁴ The remaining instances of “Standard English” and “proper English” are distributed across the island’s other urban areas. They are especially related (in nine cases) to areas that seem to be clustered in the East-West corridor (minus Port of Spain and the adjoining municipalities of Laventille and Morvant; see further section 3). Some respondents apply to the Northwest linguistic descriptors that do not involve the term “English.” Twelve of these linguistic descriptors display a social or attitudinal dimension that thematically reflect the social descriptors discussed in section 5, such as “stush talk” (AF-S-F3), “stuck up accent” (MI-N-F7), or “forced accent” (EA-N-M1). Four descriptors display a racial dimension, such as “proper white people accent” (AF-S-F14) or “white talk” (MI-S-M9). Finally, the theme of American-ness is incorporated in eight linguistic descriptors applied to the Northwest, such as in “American accent” (EA-S-F1) and “yankee talk” (AF-S-M9), and the broader theme of foreignness in six, as in “foreign accent talk” (MI-N-F3).

After “English,” the next most widespread term in linguistic descriptors is “dialect” (forty-two occurrences, twenty-two respondents), “Creole” used as a noun (thirty-one occurrences, nineteen respondents), “talk” used as a noun (thirteen occurrences, ten respondents), “twang” (three occurrences, three respondents), “jargon” (two occurrences, two respondents), “slang” (two occurrences, two respondents). These terms are part of linguistic descriptors that are generally applied to regions outside the Northwest. The term “dialect” tends to be used on its own. It is qualified only a few times as, e.g., in “Trini dialect” (EA-N-F5), “Indo-Trini dialect” (AF-N-F3), “broken dialect” (MI-S-F4). “Creole” tends to be combined with “English” (i.e., “English Creole”; six occurrences), and “bush” (i.e., “bush creole”; three occurrences). The term “talk” tends to be combined with “bush” (i.e., “bush talk”; five occurrences) and “bush Indian” (i.e., “bush Indian talk”; five occurrences). Two respondents use the descriptors “rural English” and “urban English” to refer to varieties outside the Northwest, which they adjoin to other linguistic descriptors among those already named above. The fact that these respondents, as well as a few others, concatenate some of the above named linguistic descriptors could either be an indication of distinct locally spoken varieties, or an indication of semantic equivalence between what could be hesitantly improvised linguistic descriptors (see further sections 3, 8). For example, EA-S-F4 joins “Creole,” “dialect,” and “rural English” when describing the linguistic usage of Penal, and EA-N-F2 joins “Creole,” “slang,” and “bush talk” when describing the linguistic usage of Trinidad’s southern part in general.

Some of the linguistic characterizations involving examples of usage convey perceptions of linguistic polarization between regions, in which the contrast between the Northwest and the rest generally stands out. According to AF-S-M6, “young females use LIKE between every three words” in West Moorings, a “Valley girl” feature which fits with the American attributes that some respondents ascribe to the Northwest (section 5; see further Eckert & Mendoza-Denton 2006). Examples of Creole usage are always related to areas outside the Northwest (and to a lesser extent outside San Fernando). AF-S-F4 encapsulates southern usage in the compound possessive pronouns my one (SE: mine) and she one (SE: hers). AF-N-F1 expresses the linguistic fault lines that she perceives in Trinidad by relating the SE expression how are you to its Northwest, and Creole equivalents (i.e., What up?, Wey yu dey?; see further section 8) to the Northeast and South. Other linguistic characterizations that illustrate a linguistic contrast between regions without involving examples of linguistic usage tend to focus on the aspects of intelligibility (e.g., MI-N-M2 says of residents of the Northwest that they are “clear and understandable,” and of southerners that they “make no sense”). Finally, a range of linguistic characterizations relate what seems to be perceived non-standard features to regions outside the Northwest and San Fernando while leaving the Northwest and San Fernando undescribed (perhaps implying that these areas represent the norm). This goes, for example, for AF-S-F4 who says of Arima residents that they “talk very old-timish,” and of Chaguanas residents that they have “very bad grammar.”

The data described above highlight what seems to be a perception of spatially and socially polarized language variation. “Standard English” and “proper English” are associated with the Northwest (or rather parts of it excluding Laventille and Morvant) and with higher social classes. On those maps where “Standard English” and “proper English” are not applied to the Northwest, the respondents still tend to linguistically contrast the Northwest with the rest of Trinidad by exclusively applying linguistic descriptors to the former that stress the notion of linguistic “correctness” or overt prestige. These are captured in characterizations such as “more civilized language” (MI-N-F1), as well as possibly in the social attributes discussed in section 5 of whiteness, American-ness, and social pretension (see further section 9 for a discussion of the prestige associations of these attributes). Other—variously named—varieties are associated with the rest of Trinidad. “Dialect” turns out to be the most recurrent common denominator of the linguistic labels seemingly opposed to “Standard English” or “proper English.” Out of the twenty-four maps where the linguistic labels “Standard English” and “proper English” are applied to the Northwest, varieties outside the Northwest are referred to as “dialect” (thirteen cases), “broken English” (eight cases), and “Creole” (seven cases). ⁵ This polarization of labels seems to reflect a unidimensional view of language variation whereby varieties are ranked according to their [+/-standard] features (see section 2). However, there are clear suggestions that the respondents’ perceptions are not reduced to an opposition between standard and non-standard. The fact that no linguistic label is consistently used on the twenty-four maps discussed above to refer to varieties outside of the Northwest suggests that the respondents might contrastively use linguistic labels to highlight sociolinguistic distinctions that they perceive within the non-standard part of the Trinidadian continuum. As section 8 illustrates, there are also parallel suggestions that the respondents make distinctions within the standard part of the continuum.

8. Perceived Distinctions within the Standard and Non-Standard Parts of the Continuum

The respondents seem to perceive fault lines within the non-standard part of the Trinidadian continuum. The most visible fault line within dialect/Creole is of an ethnic nature. Out of the 323 linguistic descriptors that do not feature the noun English, forty-three feature ethno-racial qualifiers. Forty of these are used by comparable numbers of Indo- and Afro-Trinidadian respondents to refer to Indo-Trinidadians. They comprise the terms “Indian” (twenty-nine occurrences), “coolie” (eight occurrences), and “Indo” (three occurrences). “Indian” mostly features in the descriptor “Bush Indian Talk” (see section 6), while “coolie” is used variously to qualify linguistic terms (e.g., “coolie jargon,” AF-S-F3; “coolie dialect,” EA-S-F3), or features in linguistic characterizations (e.g., “People speak like bush coolies,” MI-N-M1). These terms are in most cases part of linguistic descriptors applied to the southern region (in thirty-five cases), and to a lesser extent to the central region (in five cases). The locations most explicitly associated with these terms are Debe and Penal (four times), and Cedros (two times), all southern locations. References to Afro-Trinidadian ethnicity are found in the linguistic descriptors “Afro-Trini patois” (MI-N-F3), “Afro-Trini dialect” (AF-N-F3), and “Afro-based broken English” (AF-S-M6). The first two descriptors are applied to northern locations or regions: in one case to both San Juan and Arima, which are both located within the East-West corridor, an area encompassing the eastern side of Trinidad’s north coast in the other case. The third descriptor is applied to Barrackpore, another Southern location. The sum of these observations suggests that Indo-Trinidadian ethnicity is perceived as linguistically salient within the non-standard part of the Trinidadian continuum. EA-S-F2 confirms the validity of this conclusion in her linguistic characterization of the central region: the comment “speak normal as well” that she adds next to the descriptor “Indian slang” clearly suggests a perception that there is a markedly Indo-Trinidadian variety of TEC.

Attributes less frequently named than ethnicity are also used to specify linguistic varieties outside the Northwest. The distribution of these attributes sometimes seems to establish hierarchies of varieties ranging from more to less marked. The most visible manifestation of perceived linguistic unmarkedness is found where the terms “normal” (used eleven times by eight respondents) and “regular” (used three times by three other respondents) are combined with linguistic terms (e.g., “normal broken Trini language,” EA-S-F8; “regular dialect,” MI-S-F3), or are integrated in linguistic characterizations such as “normal sounding people” (EA-N-F3). The terms “normal” and “regular” are opposed to a diverse, and not necessarily consistent, set of attributes. They may be opposed to ethnically marked varieties (six cases, all referring to Indo-Trinidadian varieties), to rural varieties (three cases), while on two maps the terms “normal” and “regular” are attached to rural varieties, with the apparent suggestion that varieties spoken in more urbanized parts of Trinidad are less normal or regular. Other ways of signaling linguistic markedness include the use of the same linguistic term with different degrees of specification to refer to distinct varieties. Examples include AF-S-F3 who contrasts the socially less specified descriptor “general Trini talk” (applied to San Fernando) and the socially more specified descriptor “Bush Indian talk” (applied to surrounding rural areas); MI-N-M4 who contrasts the unspecified descriptor “dialect” (applied to the central region) with the linguistically specified descriptor “thick dialect” (applied to the South); and MI-S-F6 who contrasts the unspecified descriptor “Creole” (applied to the central region) with the linguistically specified “extra Creole” (applied to Debe and Penal). The Southern respondents tend to place less marked varieties in the South, more specifically in San Fernando (eleven and five times, respectively), followed by the central subregion (seven times). By contrast, there is a stronger tendency among the northern respondents to place less marked varieties in the North, which suggests that TEC varieties are ranked differently across the North/South divide, in both cases with a bias in favor of geographically close varieties. The bias reported here reflects the tendency noted in PD studies to rate one’s own variety as most neutral or normal (cf., Preston 2003).

There are indications that some respondents perceive different sorts of “Standard/proper English.” The main category of distinction involves an opposition between “American” and “local,” which in most cases is summarized in terms of differences in accent. For example, EA-S-F1 observes that residents of West Moorings speak a “very proper English with an American accent,” while those of San Fernando by contrast speak an unspecified “proper English.” Elsewhere, a “flat accent” (EA-N-F4) or “Trini accent” (MI-S-F6) is opposed to the American character of Northwestern Standard English usage. Some respondents perceive degrees of standardness. For example, AF-S-M9 describes the English of Port of Spain as “very proper,” and that of San Fernando as “slightly less proper,” while EA-N-F2 contrasts “proper English” (associated with Port of Spain) with what she describes as “standard English kinda” (associated with Arima in the East-West corridor). The respondents may articulate their perceptions of linguistic hierarchies by simultaneously evaluating the levels of standardness and non-standardness attached to a given variety. For example, EA-N-F5 sees a contrast between the Northwest, which she associates with Standard English, and the adjoining northern region whose residents in her opinion speak “in between Standard English and dialect: probably mesolect” (see further section 9). Some linguistic descriptors raise doubts as to where some respondents see a boundary between Standard English and other varieties. An example of unclear boundaries is AF-S-M1, who contrastively uses the descriptors “stush English” (applied to the Northwest), “rural English” (applied to the eastern subregion), and “Creole English” (applied to San Fernando). The use of these descriptors raises the question whether the respondent sees the former two varieties as commonly deprived of Creole features and thus opposable to San Fernando’s variety, which perhaps is the only of the three that exhibits Creole features in the respondent’s view.

In sum, this section highlighted that some non-standard/Creole varieties stand out in the respondents’ perceptions, especially those varieties spoken by Indo-Trinidadians. At the same time, the respondents tend to make a distinction between “Americanized” and “local” or unmarked varieties of SE within the standard part of the Trinidadian continuum. These are indications that linguistic variation is not only perceived in unidimensional terms of conformity to the standard, but also along other dimensions defined by, most visibly, ethnicity and foreignness/localness.

9. Characterizations of Linguistic Diversity at a Local Level: Co-Existing Sociolects and Repertoires

Trinidadian studies summarized in section 3 imply that SE and TEC can generally be found side by side in the Trinidadian speech community, even if they acknowledge that the boundary between the two may be somewhat blurred. If they perceive it, how do the respondents represent this cohabitation? Respondents appear to associate 64.8 percent of locations/areas to which linguistic descriptors are applied with one single variety (in 18.1 percent of the cases described as “standard,” “proper,” or “stush” and related to the Northwest; see section 6). A restrictive interpretation of this tendency could be that it manifests a perception of a linguistic landscape of which most parts are regarded as largely monodialectal. By contrast, the practice pointed out in sections 4.1 and 7 of juxtaposing linguistic descriptors could point to perceptions of linguistic diversity at a local level. The possibility of such perceptions is borne out by a limited number of maps that display characterizations of ranges of varieties present at given locations. For example, when observing that “much more bush Indian type [is] heard over anything else” in the South, EA-M-N3 does not preclude the possibility that other varieties may occasionally be heard locally. Other maps provide characterizations that imply perceptions of co-habiting group-specific lects or ways of speaking. For example, AF-S-M1 observes that San Fernando residents have “different accents based on class,” and that in more rural Southern locations “Creole might have different accents based on race.” AF-S-F8 describes the Northwest as a region where “many use accents,” thus leaving open the possibility that some locals may not have “accents,” and EA-S-F1 specifically relates “high pitches” to women in the central region, thus ascribing one salient Indo-Trinidadian stereotype—studied in detail by Leung and Deuber (2014)—to one specific social category.

Perceptions of linguistic diversity are also conveyed by characterizations of what the respondents seem to regard as average local linguistic repertoires. The most unmistakable characterizations of linguistic repertoires include references to practices of language alternation, that is, a conversational pattern of code-switching whereby two language or language varieties are kept distinct in conversation, generally for indexing changes in situational settings (Auer 1999). While describing linguistic usage in the East-West Corridor, AF-S-F1 comments that locals “mostly [use] standard English with a mixture of Creole for talking to close friends.” This observation implies that SE forms a default variety in the concerned area, from which individuals may switch into non-standard varieties for the specific purpose of indexing informality. Other descriptions of code-switching refer to audiences, which the respondents see as situationally determining individuals’ linguistic choices. For example, AF-S-M5 describes linguistic usage in San Fernando in the following terms: “Speech changes due to audience/based on who person is interacting with (raw Creole/proper English).” MI-S-F1 suggests that the typical resident of Trinidad’s Eastern coastline “speaks broken English but knows when to speak Standard English” (Figure 3), suggesting that standard and non-standard varieties form distinct components of average local stylistic repertoires, each of which is deployed for performing specific situational functions. By contrast, some characterizations suggest that residents of certain areas are less able to code-switch. For example, EA-N-F3 remarks of North Coast residents and Southerners that there is “no variation in their accent,” although this observation may as well convey a perception of linguistic homogeneity at a local level.

A specific category of linguistic descriptors revolves around the theme of “mixture.” That theme is apparent in thirty-three descriptors in the form of, for example, “mixed speech” (AF-S-F11), “mix of standard and Creole” (EA-N-M8), “mixture of mainly local dialect with a pinch of standard English” (EA-N-F4). The respondents may use imageries of linguistic mixture to refer to perceived transitional stages between two regional varieties or between a variety of TEC and SE. However, these characterizations may as well constitute folk descriptions of language mixing, that is, a conversational pattern of code-switching whereby two varieties are mutually imbricated (Auer 1999), and which Winford (1997) and Youssef (1993; 1996) describe as a defining feature of Trinidad’s linguistic landscape (section 3). In most cases (thirty-one out of thirty-three), the respondents describe “mixtures” as involving SE on the one hand, and, on the other hand, “Creole,” “ghetto,” “slang,” “accents,” and “words.” In two cases, “mixtures” also involve language varieties unrelated to English, namely “French patois” in the northeastern subregion (AF-S-M3) and “Hindi” in the Southern region (AF-S-F8). The respondents locate mixtures of SE and other (TEC) varieties mostly in urban areas (in twenty-nine cases out of thirty-one), more specifically in Port of Spain, followed by San Fernando, the central region, and the East-West corridor. Another suggestion that the respondents might be describing language mixing practices when using the imagery of “mixture” is found where they describe ways of speaking in terms of the respective sizes of their standard and non-standard components as if these were consistently found imbricated in one another. For example, what AF-S-F10 calls “regular Creole with a definite mix of standard” in her description of linguistic usage in Point Fortin (a southern location) could refer to a pattern of language mixing of which the dominant linguistic component consists of non-standard features. Finally, some respondents are less specific on what forms the dominant linguistic component of the mixture, such as EA-N-F4, who describes northern usage as “Standard English combined with dialect to form sentences.”

In the limited number of cases where they acknowledge linguistic diversity at a local level, the respondents rationalize it as sociolectal variation, or as alternation between standard and non-standard varieties in repertoires. One must be cautious when interpreting characterizations that involve the notion of “mixture.” While some may refer to language mixing, others may just aim to convey a perception of gradience illustrated by a range of PD studies using degree-of-difference tasks (e.g., Kuiper 1999; Benson 2003; see further Canut 2002).

10. Discussion

This study elicited spatially and socially contextualized folk perceptions of sociolinguistic distinctions. The respondents generally ordered their perceptions around a twofold spatial distinction between North and South (with scope for a secondary intermediate region), and between Trinidad’s main two cities and the rest. These distinctions coincide to some extent with degrees of urbanization and colonial legacies of ethnoracial distribution. Most of Trinidad’s population is concentrated in the North, where Afro-Trinidadians have historically been concentrated. By contrast, Indo-Trinidadians have been dominant in Trinidad’s more rural southern part since the time of Indian immigration. The respondents tend to display sociolinguistic awareness of Trinidad’s small white population, which they locate in the Northwest, Trinidad’s historic center of power. The North/South divide is linguistically expressed in the tendency among respondents to reserve the descriptor “Indian” for characterizing Southern varieties. Interestingly, Afro-Trinidadian ethnicity is hardly ever referred to in descriptors applied to Northern varieties, whose most visible distinctive feature is that they tend to include one variety labeled as “ghetto,” associated with the Northwestern municipalities of Laventille and Morvant. The maps tend to highlight an urban-rural divide opposing Port of Spain and suburbs, as well as the southern city of San Fernando, to the rest of the island. The highlighting of this divide serves to spatially contextualize the distinction that the respondents make between SE and other variously labeled varieties perceptually distinct from SE. Apart from being associated with urbanity in general, SE also tends to be described as an attribute of whiteness and American-ness, which the respondents specifically locate in the Northwest. Where the descriptor “Standard English” is applied to varieties outside the Northwest, it is not conflated with whiteness, American-ness, nor with any ethnoracial attribute, and thus seems to refer to a less marked, and more local, variety of SE. The respondents showed bias with respect to the range of varieties that they described in the northern and southern regions, with the southern respondents more likely to make more sociolinguistic distinctions in the latter.

The recurrent use of polarized linguistic descriptors could generally support an analysis of Trinidad’s linguistic landscape as one in which only two linguistic systems or varieties co-exist while suggesting that these linguistic systems or varieties are spatially and socially bounded. It also suggests that variation tends to be viewed unidimensionally in terms of distribution of [+/-standard] features. On the one hand, SE is to a large extent associated with whites and urban middle and upper classes. On the other hand, “dialect” and “Creole” are the terms most often applied to varieties outside the Northwest, which supports Mühleisen’s (2001) earlier observation that TEC—despite inconsistencies in its labeling and the occasional use of the label “broken English” on the maps—is increasingly perceived as being a linguistic system or variety of its own intrinsically opposed to SE (Mühleisen 2001; section 3). Section 3 set out Winford’s (1997) view that contact between acrolect and basilect in Trinidad has not given rise to a mesolect, and can instead only be observed in the form of code-switching between two clearly marked out varieties. Some comments made by the respondents lend support to this view by suggesting that SE and non-standard varieties form situationally distinct components of individual repertoires, at least in rural areas (section 9). More generally, the fact that respondents tend to define varieties in terms of their SE component in their linguistic characterizations that involve the notion of “mixture” might imply that the use of SE remains conversationally salient, an indication that it forms part of a pattern of code-switching whereby the participating varieties tend to be kept conversationally separate (cf., Auer 1999). Some patterns in the respondents’ perceptions must be carefully contextualized for them to be seen as compatible with Winford’s (1997) dual system perspective. In order to conform to that perspective, the North/South divide that most respondents perceive should not be analyzed as a boundary between two distinctive linguistic systems, but rather as the outcome of the respondents iconizing certain suprasegmental and lexical features of southern varieties, such as high pitches or Hindi words that make them sound distinctive (section 8).

Still, it remains possible to reconcile the respondents’ perceptions with other perspectives. Some respondents’ perceptions seem compatible with assuming more than one linguistic boundary in Trinidad. The respondents tend to describe varieties specific to the rural South, generally ascribed to Indo-Trinidadians, as well as implicitly portrayed as furthest removed from SE, as the descriptors “extra Creole” or “thick dialect” specifically applied to them suggest (section 8). The perceived linguistic boundary between these ethnic southern varieties and those unspecifically labeled “dialect” or “Creole” seems to reflect not only historical patterns of ethnic concentration, but also historical levels of exposure to SE. As the ethnic group that urbanized last, Indo-Trinidadians gained relatively late exposure to SE and to the Barbadian precursors of TEC that spread from Trinidad’s historic urban centers, in and around which most of the Afro-Trinidadian population was already concentrated by the late nineteenth century (section 3). The regions adjoining these historic urban centers are, therefore, where one could expect an intermediate variety with a mesolectal character in keeping with scenarios that describe decreolization as first occurring in urban areas (Rickford 1987; Romaine 1988). One suggestion that these regions are perceived as being home to varieties intermediate between SE and rural southern varieties could be found in the respondents’ tendency to associate with them “mixtures” of SE and dialect or Creole. “Mixture” could refer to stable intermediate grammatical systems. More probably, it could refer to a pattern of language mixing—which Youssef (1996) claimed is becoming a norm in Trinidad. Language mixing has been shown to sometimes give rise to “fused lects” as the structural saliency of two given interacting languages or varieties recedes (Auer 1999). Therefore, both interpretations of the folk notion of “mixture” could warrant a subdivision of Trinidad’s linguistic landscape into (at least) three linguistic areas as opposed to just two.

11. Conclusion

The spatially and socially contextualized sociolinguistic perceptions that were elicited in this study consistently oppose standard and non-standard varieties. The former are referred to as “Standard English” or “proper English,” and are predominantly associated with the historic urban centers, while the latter are referred to as “dialect,” “Creole,” or “broken English,” and are predominantly associated with rural regions and regions adjoining the historic urban centers. The respondents seem to make distinctions among non-standard varieties: some are urban and described as “ghetto” or as involving “mixtures,” while some are rural and categorized according to whether they display “Indian” attributes. Meanwhile, the respondents tend to distinguish between a “white,” “Americanized,” or “foreign” variety of Standard English—strongly associated with the Northwest—and a more local one, or at least one that is unmarked. If taken at face value, these observations suggest that perceptions of variation in Trinidad are ordered along two dimensions, namely, standardness and ethnicity. Besides, the recurrence of the theme of “mixture” in the linguistic descriptors used by the respondents suggests the possibility of establishing a variety located at an intermediate position between Standard English and TEC if one considers the possibility that “mixture” could be a folk term referring to language mixing or to “fused lects” (cf., Auer 1999). Although they offer limited insights into actual linguistic patterns, the data presented in this study provide an outline of salient social fault lines, which most visibly coincide with the urban-rural divide and ethnicity. It is hoped that future sociolinguistic studies of Trinidadian language data will test the linguistic significance of these social fault lines while seeking to establish linguistic boundaries in the Trinidadian continuum. It is also generally hoped that more sociolinguistic studies will acknowledge the merits of involving folk perceptions in their design, especially when these studies aim to establish the spatial dimensions of patterns of language variation.