Competing Reflexive Models of Regional Speech in Northern Ohio

Introduction

The notion of enregisterment (Agha 2003, 2007) refers to the processes through which sets of language forms become invested with social meaning and, more basically, become understood socially as coherent sets of forms. Campbell-Kibler (2012) argued that among Ohio speakers, the boundary between white, nonrural Inland North and white, nonrural Midland speakers is enregistered, but exists within a limited domain of recognition, is unstigmatized, and is frequently constructed as idiosyncratic. This value-neutral treatment contrasted sharply with mentions of the widely known and stigmatized varieties of Southern and rural speakers and racialized images of urban youth “slang.” That work, however, was based on a population of largely central Ohioans. In the current paper, based on data collected at the Great Lakes Science Center in Cleveland, Ohio, we ask how northern—primarily northeastern—Ohioans view the sociolinguistic landscape of Ohio, with an emphasis on how they agree or disagree with the perspectives of the mostly central Ohioans documented in Campbell-Kibler (2012). In so doing, we seek to understand how differently placed speakers understand and engage with processes of enregisterment.

Systems of language ideology in the United States assign social weight to the status of “having an accent” as contrasted with “not having an accent,” regardless of the accent itself (Bonfiglio 2002; Lippi-Green 2012). As a result, awareness of a difference between two “unaccented” populations has potentially serious implications, as it raises the possibility that either or both sets of speakers may become defined as “accented.” In Campbell-Kibler (2012), the primarily central Ohioan speakers mentioning the Midland/Inland North boundary consistently positioned the north as the differing variety, in contrast to a central variety positioned as neutral or standard. This finding raises the question of whether northerners share the perspective that it is their variety which is marked and, if they do not, whether they instead show less awareness of the Inland North/Midland divide or they position their own variety as normative.

The results presented below suggest that they are split on this issue, with subgroups ignoring the Midland/Inland North difference, marking their own speech as distinctive and characterizing the Midland as divergent. The presence of this last group in northern but not central Ohio shows the importance of personal stakes in the reflexive models (Agha 2007) speakers circulate and underlines the ideological importance of being linguistically unremarkable.

Connecting the Linguistic and the Social

One of the enduring sociolinguistic puzzles is the nature of the processes which connect individual elements of language on the one hand and social elements such as places, situations, and groups on the other. These processes include the negotiation of social meanings, theorized by Silverstein (2003) as the accumulation of indexical orders through contextually grounded creative uses of existing indexical links. But they also include the (re)construction of the imagined linguistic objects themselves: the classification of forms into coherent sets capable of meaning. Agha (2003, 2007) has proposed an understanding of these processes under the term of linguistic “enregisterment.” Specifically, he suggests that human interaction is undergirded by reflexive models through which people make sense of their own and others’ behavior. Enregisterment refers to the processes through which models are formed, linking specific material acts to each other and to constructs like person types and social actions.

The current paper is concerned with what Agha refers to as a speech chain network, which is the network created by a series of speech events through which a given cultural message is spread through social space. Receivers of the cultural message may in turn become senders in new speech events, thereby broadening the set of speakers capable of, in the case of enregisterment, recognizing linguistic forms involved in a register and associating them with particular social content. He emphasizes the importance of role alignment in this process, through which speakers exposed to metapragmatic discourse about forms are prompted to align themselves in some way (not necessarily in a positive or affiliative way) with a characterological figure in such discourse (Agha 2007:203). The roles available to be aligned with and the alignment that results have a strong impact on the uptake over time of such circulating objects. This is because such models create the sociolinguistic spaces available for speakers to inhabit, thereby shaping their sociolinguistic worlds. Different speakers, then, are likely to engage with the same models in very different ways, based on how they are able to align with those models to construct their own place in the sociolinguistic world.

Building on Agha’s (2003, 2007) and Silverstein’s (2003) insights, Johnstone, Andrus and Danielson (2006) have proposed a more specific model through which sets of language forms may come to be understood as belonging to a specific place, using the example of Pittsburghese. In the case of Pittsburghese, features most used by the working class in the area became noted as a result of economically triggered diaspora. As Pittsburgh exiles encountered others who remarked on their accents, features previously linked to general meanings of normative speech took on new meanings, or n+1 indexical orders (Silverstein 2003), largely related to class. These new meanings in turn became available not only for presentations of “authentic” or straightforwardly claimed local identity through simple use, but for performatively marked citation, particularly by those for whom the forms were not part of their day-to-day speech patterns. The use of key lexical items such as yinz and phonetic features, most centrally the monophthongization of /aw/ as in the frequently cited dahntahn, are thus used to index personae linked to an authenticated, though not always valorized, image of the canonical Pittsburgh native.

Johnstone, Andrus and Danielson (2006:83) suggest that the first stage of the development of Pittsburghese is one in which the soon-to-be-marked variants (e.g., /aw/ monohphthongization) are “not invested with social meaning,” because speakers are not yet aware of the forms’ distinctiveness from other forms (e.g., diphthongal /aw/), nor their correlation with specific demographic groups (namely, young, working-class males). But we suggest that it is not the case that forms such as yinz or dahntahn lack meaning prior to the mobility-triggered acts of noticing which led to them being imbued with the particular meanings documented. Rather, we argue that they likely existed within a larger register, with meanings such as normative, unremarkable, or (in the right contexts) American, non-Southern, and native English.

In the current paper we show that northern Ohioans are more likely than other, mostly central, Ohioans to offer perceptual dialectological views which lack the constructs of the northern or Cleveland accent. We argue that this preference stems not only from the negative evaluation of having marked or accented speech, but from the positive evaluation of unmarked, normative speech, particularly in the context of U.S. language ideologies more generally. Thus, the alternative to a model in which yinz indexes Pittsburgh or raised trap/bath ¹ indexes Cleveland is not one in which they index nothing. Rather, it is one in which they both, together with you, lower trap/bath, high front tokens of fleece, the lexical item cat, and so on, index normative, unremarkable American English. A given register’s treatment of specific forms is in competition with those of other models, which themselves assign form-meaning associations that speakers/hearers may be reluctant to relinquish.

Variation and Language Beliefs in Ohio

Ohio is a particularly fruitful location for research on the dynamics of accent due to its position in both the actual and imagined dialectology of the United States. The northern portions of Ohio fall in the dialect area known as the Inland North (Labov, Ash & Boberg 2006), Upper North (Carver 1987), or North (Frazer 1996), most prominently known for participation in the Northern Cities Shift (NCS), in which the trap/bath vowel is raised, fronted and/or diphthongized, while the lot vowel is fronted, and the thought vowel is lowered and/or raised, avoiding the cot/caught merger found elsewhere (Eckert 2000; Gordon 2001; Thomas 2001). NCS features also involve the vowels dress and kit, reported as backing (Eckert 2000) or lowering (Gordon 2001), while strut backs (Eckert 2000; Gordon 2001). Central Ohio is part of the Midland (Labov, Ash & Boberg 2006; Frazer 1996) or Lower North (Carver 1987) dialect region, whose most well-studied feature, shared with the West, Canada, and eastern New England, is the cot/caught merger-in-progress, in which the lot and thought vowels become indistinguishable (Labov, Ash & Boberg 2006), but is also a leading area for the fronting of goose and goat (Labov, Ash & Boberg 2006). In Columbus specifically, younger speakers have been noted backing trap/bath, pulling their productions further away from the Inland North productions (Durian 2012).

Developed due to migration patterns originally (Wilhelm & Noble 1996), the Midland/Inland North boundary within Ohio appears to have remained remarkably stable over the years (Thomas 2010). Some parts of Ohio show patterns associated with the Inland South (Labov, Ash & Boberg 2006) or Upper South (Carver 1987) dialect region, although it remains a matter of debate exactly how widespread these patterns are within the state. Flanigan (2000, 2005) has argued that elements of the South Midland extend at least through the southern third of Ohio and possibly beyond. Ohio also features parts of the Western Pennsylvania variety in the eastern areas of the state (Labov, Ash & Boberg 2006), and some Cincinnati-specific attributes have been documented (Boberg & Strassel 2000). This leaves Ohio with a great deal of regional dialectological complexity, in addition to language variation associated with racial, ethnic, and gendered variation as well as nonnative English varieties.

Work in Ohioans’ perceptual dialectology has been less common than work in Ohio dialectology. Ohio is typically represented in countrywide map tasks by others in one of two ways: as a neutral, standard area (Preston 2002) or as a haven for rural speech (Fought 2002). Two papers have explored Ohioans’ perceptions of variation within the state. Benson (2005) collected data from twelve speakers from around the state, using a map task and a degree of difference rating task of Ohio and nearby areas. While she discusses within-state differences, it is difficult to interpret them with confidence, since each area is represented by only a few speakers. However, key themes emerge across her data, including an overall north/south divide, distinct corners for northeast and northwest Ohio, and repeated mentions of rural and Southern ways of speaking.

The other existing perceptual dialectology study on Ohio, Campbell-Kibler (2012), had visitors to the Center of Science and Industry in Columbus, Ohio, read word lists and complete perceptual dialectology tasks. A total of 152 museum visitors were interviewed, primarily from central Ohio. This work found a clear difference between, on the one hand, widely circulated accent constructs associated with Southern and/or rural speakers and with urban African American speakers, and, on the other, a contested, less well known accent associated with otherwise socially unmarked speakers of northern Ohio. All of these were contrasted against the ideologically unaccented, unmarked speakers of central Ohio, who are implied by contrast to be nonrural, nonurban, and white. Otherwise unmarked northern speech was described in neutral or lightly negative terms like “silly” and explicitly marked as unimportant. Despite these mitigating efforts, the directionality of this process was consistent, with all but one speaker (a Clevelander) positioning the central variety as the unaccented norm. The current research extends that work by collecting similar data from a largely northern population. We ask to what extent northern Ohioans share their neighbors’ view of Ohio-based sociolinguistic variation generally, with a particular emphasis on how the Inland North/Midland boundary is or is not constructed.

The terms partially versus fully enregistered, used in Campbell-Kibler (2012), imply an unfortunate finality in the notion of enregisterment which is at odds with the fluidity and ongoing nature of the processes. Drawing on Agha’s dimensions of register organization (Agha 2007:169), we might more usefully reframe the conclusions from Campbell-Kibler (2012) to suggest that the register(s) tied to Cleveland or northern Ohio by those participants differed from the other registers mentioned primarily in lacking images or persona types (e.g., “hillbilly” on the one hand and “Walter Cronkite” on the other) and showing less strong values.

In addition, Campbell-Kibler (2012) argued that participants who did mention northern speech as distinct expected it to be less familiar to the interviewers than they did the other speech types they mentioned. Reports of northern speech were repeatedly supported with reference to direct personal experience with northern speakers, an evidential strategy rarely used for Southern or rural speech or racialized urban “slang,” as termed by participants. We might envision this as meta-information regarding domain of recognition; speakers conceptualize the register as being of relatively limited circulation, such that a stranger might need convincing of its existence. In the current paper, we compare new data from northern Ohioans to the other Ohioans from that study to ask how a different regional perspective, and therefore personal investment, influences the circulation of reflexive models.

Methods

Data for this study were collected at the Great Lakes Science Center, in downtown Cleveland, in northeastern Ohio. Museum visitors were approached in the lobby and asked if they were willing to participate in a study on Ohio dialects and whether they had lived in Ohio their entire lives. Those who had were introduced more fully to the goals and procedures of the study and asked for their consent. Those who agreed were taken to the main research area, a table set up in a hallway near the museum entrance. They were audio recorded throughout the ten-minute interview, which consisted of three word lists (spaced throughout the other tasks), a residential history, an open-ended verbal guise task, and a perceptual dialectology task using a blank map of Ohio.

The map used in the map task (shown in Figure 1) was the same as in Campbell-Kibler (2012), showing Ohio with only small portions of surrounding states and a few major cities labeled. Participants were given the map and asked to indicate the “different places people talk the same way.” They were not asked to write labels for the areas on the map directly, since the previous study had found that participants did not do so even when asked. After they had completed the map, participants were asked by the interviewer to explain their maps, and prompted to continue with general prompts such as “Anything else?” but not directed to specific regions to explain them. After a first explanation, participants were then asked what specific language qualities they had based their judgments on, if any. Interviewers marked on small duplicate maps which regions were indicated by deictic references to the map, to facilitate later analysis of transcripts.

OPEN IN VIEWER

Figure 1.

Map Used in Map Task.

In total, sixty-three participants were interviewed, with an average age of thirty-eight years (standard deviation fifteen years). Due to the short, anonymous nature of the interviews, no questions were asked regarding race and socioeconomic status. The interviewers noted estimates of race, classifying fifty-two of the participants as white, six as African American, and failing to classify two more with confidence, and gender, classifying thirty-seven as female and twenty-three as male. Such classification does not, of course, reflect the complexity of lived experiences of race or gender, but for the purposes of the current study, captures these constructs only slightly less effectively than a short-answer question response. Regional backgrounds were classified on the basis of the regional histories, classifying as northern speakers who had lived their lives even with or north of Mansfield, which corresponds roughly to the boundary documented by Thomas (2010). Forty-four of the participants were from northern Ohio (twenty-nine from the Cleveland area), six were from elsewhere in Ohio, and six more were originally from northern Ohio but had since moved elsewhere. The remainder had more complicated regional histories or, despite prescreening, had lived more than five years outside Ohio and are not included in the current discussion.

To maximize the generalizability of the analysis, the discussion below will combine data from these participants with those discussed in Campbell-Kibler (2012), interviewed using similar techniques at the Center of Science and Industry in Columbus, Ohio. Due to methodological differences between the studies, the data will be joined in two pools: one for the quantitative analysis of commentary and one of the maps themselves. For the former, the data from the Cleveland participants are combined with the data from the seventy-one participants (eleven from northern Ohio and sixty from elsewhere in Ohio) in the central Ohio dataset for whom verbal commentary was available. Thus the percentages reported below are based on a corpus of sixty-six nonnortherners and fifty-five northerners. The six former northerners are included in the map analysis, which is not broken down by region, but due to their small number, are not included in the discussion of verbal commentary.

Verbal commentary was analyzed by noting common themes and variation in how the same or similar groups or places were portrayed. Participants were coded quantitatively for mentions of specific stances, including whether they mentioned a distinct northern Ohio region, southernness as a construct within Ohio, rural speech as a construct, and standard or unaccented speech as a construct. All of these were coded purely by mention (yes/no). In addition, urban speech was quantitatively coded for mention and for characterization (neutral, standard, or with any of the constructs of race, youth, or slang). Determinations of significant difference in commentary are based on logistic regression with responses as dependent variable and regional origin as predictor.

The map analysis is based on an overlapping but not identical set of participants: sixty-six participants from both datasets (twenty-six from the north, nineteen from elsewhere, eight who moved from the north to elsewhere in the state, and thirteen with other complex regional histories or who declined to state their regional history) who completed map tasks by delineating regions. The remaining sixty-six participants who also were asked to complete map tasks (thirty-five at the Great Lakes Science Center and thirty-one at the Center of Science and Industry) either responded entirely verbally or marked their maps in ways that did not delineate regions in the style of dialect maps (e.g., underlining city names or drawing lines between cities to indicate commonality).

Given the complexity and variability in hand-drawn maps, reliable quantitative analyses are challenging and a range of digital analysis techniques have been explored (Montgomery 2007; Cramer 2010; Evans 2011, 2013). These have typically been aimed at creating a variety of composite maps (e.g., all regions, 50 percent agreement, and regions with a specific label attached) and have been quite successful.

One technique is to combine areas described by the same or similar conceptual terms (Evans 2011), which is useful for revealing the mapping of concepts such as “country” onto geographic space. In our case, the divisions of interest are regional rather than conceptual. It is unusual for participants to indicate precisely matched regions, so to explore which regions are commonly identified is typically accomplished via visual inspection of composite maps, locating areas of dense representation. As the composite maps below in Figure 5 show, however, the dense overlapping of our maps makes this challenging. We present a technique for identifying clusters of geographically similar regions quantitatively in order to separate and examine clusters. We identify the geometric center of each hand-drawn region using QGIS (Quantum GIS Development Team 2012)—an open-source Geographic Information Software (GIS) package—then use k-means clustering to identify clusters of these centers. We take each cluster to indicate a perceptual dialect region circulating among the population.

The hand-drawn maps from both data sites were scanned, and Adobe Photoshop was used to digitally encode the lines drawn by participants. The resulting images were geo-linked and converted to vector data using QGIS. In order to perform the clustering algorithm, it was necessary to reduce each region to a single point. The geometric center was selected as the most consistent technique for assessing a core tendency or area meant to be captured by the region by the participant. It is not always the case that the geometric center will coincide precisely with the conceptual center (if, indeed, there is one) intended by the participant. For example, a region centered around Cleveland will necessarily have a geometric center somewhat to the south, given Cleveland’s positioning on the northern border of the available territory of the map. Nonetheless, the geometric center indicates to a certain extent the central tendency of the region. We hypothesize that regions with similar centers will be of similar shapes. This is, of course, not mathematically necessary, but, to the degree that it holds, it will support the cluster model as an approach for identifying shared geographic conceptualizations.

The raw data for this analysis were the individual regions delineated on participant maps. Some markings can be ambiguous: is an east-west line marking a northern region, a southern region or both? To address this, we used the verbal commentary to determine region status. All regions named or referred to in any way by the participant during the map explanation were included, even those invoked only in passing. Thus, an east-west line might yield only one region if the participant explained that the southern portion was distinctive, but would yield two if they remarked that the southern speakers were different from the north or that the south and north were different from one another.

The centroid of each such region was calculated and the resulting points were analyzed using k-means cluster analysis in R (R Development Core Team 2007) to identify clusters of centroids most similar to each other. A scree plot was used to establish the optimal number of clusters (seven), and the resulting clusters were used to create smaller composite maps. This procedure was followed for the pooled data, as well as subsets based on responses only from northern participants and from nonnorthern participants. These yielded similar structures, so the pooled analysis is presented here. The composite maps for the resulting clusters are presented below.

The next section presents a statewide analysis of the hand-drawn maps using this technique, as well as an overall examination of regional differences using overall composite maps. From there, we turn to an analysis of the verbal commentary, documenting common themes and the differing reactions to the Cleveland or northern accent. Finally, we discuss the implications of these results for our understanding of enregisterment, arguing for the meaningfulness of normative or unremarkable forms.

Map Results

While individual participants show variability in their conceptualizations of Ohio, the basic structure of regional division proved to be stable across the regional background of the participants. Accordingly, in this section we sketch out this basic structure based on combined data from both museums. These are presented in Figures 2–4. Each figure also includes the number of individual maps whose regions are included in the cluster. Note that a given cluster may have more individually drawn regions in it than maps included due to participants marking more than one area. For example, a participant who delineated both Cleveland and the northeast generally, or Cincinnati and the southeast generally, would contribute two regions to the relevant cluster. No cluster includes more than three such repeated instances.

OPEN IN VIEWER

Figure 2.

Southwest (37 Maps) and Southeast (33 Maps) Clusters.

OPEN IN VIEWER

Figure 3.

Small Northwest (12 Maps), Large Northwest (3 Maps) and Northeast (16 Maps) Clusters.

OPEN IN VIEWER

Figure 4.

North (19 Maps) and North-Central (13 Maps) Clusters.

The first pattern visible in the clusters is the importance of the north-south division. Ohioans share an orientation to language variation as running north to south, with east-to-west distinctions occurring within the north/south strata.

Despite the two southern clusters shown in Figure 2 being visibly divisible into a southwest set centered on Cincinnati (left) and a southeast cluster centered on the Appalachian areas of the West Virginia border (right), many individual regions span the entire state east to west. No maps were drawn marking entire eastern versus western regions. The conceptualization of Ohio as divided north to south echoes that of the United States as a whole (Preston 1986, 1997) and, as we will see below, is explicitly linked by speakers to that larger division. This is perhaps a result of Ohio’s geographic positioning and/or its dialect boundaries themselves. Note that, for example, Evans (2011) has documented that residents of Washington State are similarly willing to subdivide their state but they orient strongly to an east versus west system of understanding state linguistic difference. Similarly, Bucholtz et al. (2007) found a strong north versus south division within California, but one which has no resonance with larger north/south U.S. discourses.

Also notable among these clusters of dialect regions is the greater diversity of the northern regions (five clusters) relative to the south (two clusters). Given the strong representation of northern and central speakers in the data, it is impossible to tell whether this bias is statewide, or whether southern Ohio participants would have subdivided the southern areas more finely. Our results contrasting the north and central Ohioans suggest that southerners are likely to represent yet a new perspective on within-state variation.

The remaining five clusters are concentrated in the north. Three clusters, shown in Figure 3, reflect east versus west differences, one for the northeast and two for the northwest, one large and one small. The small northwest region (far left) appears to be primarily Toledo-centered, while the larger set (center) comprises roughly the northwestern quarter of the state. While it’s possible that these apparently redundant northwest clusters reflect substantive disagreement about the linguistic nature of northwest Ohio, the small number of maps in the larger cluster make it impossible to tell.

The two remaining clusters, shown in Figure 4, include the entire west-to-east span of the north but differ in the depth to which they assign the northern area, either only a band across the top of the state (left) or a more substantial portion of the state (right). Note that no pattern is seen linking this difference to region of origin of the map drawer. This is potentially due to the fact that these lines were sometimes the only line of the map, representing a two-way north/south division, and were at other times the more northern of two lines representing a north/central/south division.

No cluster was found indicating a set of “central Ohio” regions. A few central-indicating regions were classified with the north-central set, suggesting that one reason for this lack lies in the centroid calculations, through which large northern regions and small but central regions are seen as more similar than they might otherwise be. Another reason appears to be the numerical scarcity of truly central regions in the data. The reason for this scarcity appears to differ for northern and nonnorthern Ohioans. Figure 5 shows composites of all the regions labeled or mentioned by northern Ohioans (left) and all other Ohioans ² (right). These two maps include the regions presented in Figures 2–4, divided by the regional background of the map-drawer rather than by location of the region. Note that not all regions are presented here, only those from participants having lived solely in either central and southern or northern Ohio.

OPEN IN VIEWER

Figure 5.

Composite Maps of All Selected Regions by Northern Ohioans (left) and Other Ohioans (right).

Comparing the two maps, we see a lightening in the central area of the “other Ohioans” map, indicating that these participants, primarily central Ohioans, were less likely than northern Ohioans to delineate and then refer to areas in central Ohio, even in passing, as they completed their maps. Recall that regions were included in the analysis if any mention, even as a point of contrast, was made verbally. This shaping and positioning of this lighter area suggests the faith that central Ohioans have in a central Ohio region. The relative lack of mention of central Ohio, compared to the northerners’ discussions, indicate the characterization of the region by central Ohioans as linguistically neutral and therefore unremarkable. Indeed, when they do comment on this area, it is to explain its lack of accent. In contrast, the northern Ohioans show a much more tenuous sense of unremarkable speech, with a more northerly area around Mansfield showing only a slight lightening. ³

This alternative view of the map data explains the lack of a central cluster across the board and recontextualizes the lack of regional difference in the clustering results. Northern Ohioans lack a “central Ohio” cluster because they lack, by and large, a notion of central Ohio as a linguistic place. They are neither likely to highlight it nor avoid it, carving it up instead into the broader north/south and sometimes east/west distinctions. In contrast, nonnorthern (mostly central) Ohioans have a strong notion of central Ohio as a place, but lack a “central Ohio” cluster because in their view, central Ohio need not be commented on. Their commentary when they do mention the area suggests that this is because in their view central Ohio’s speech is best described as a lacking an accent.

Common Themes in Verbal Commentary

While the map analysis can help us understand the geographic components of the imagined sociolinguistic landscape of Ohio, it is the verbal commentary that allows us to understand the ideological import of these areas. In particular, the differing treatment of central Ohio by the two regional groups raises questions about whether they differ merely in their geographic detail of their ideological picture or at a more fundamental level. Do the two groups agree on what the dominant elements are of language variation and how they are linked to sociogeographic character? We find that they do, by and large. Participants statewide show a belief in the existence of an unmarked standard speech, while focusing heavily on the north versus south and urban versus rural oppositions in conceptualizing speech variation. Only three differences are noted: in addition to the issue of central Ohio noted above, northern Ohioans may show a more positive sociolinguistic association with urban areas, and they show a very different conception of northern and particularly northeast Ohio.

A common theme is the existence of nonaccented speech, best displayed by the central Ohioan in (1), taken from Campbell-Kibler (2012). While Walter Cronkite offers more of an explicit guide as to possible features found in this variety, both the descriptor normal and the claim of God as a cospeaker point to an intense ideology of legitimacy, documented thoroughly by Lippi-Green (2012) under the phrase “myth of the non-accent.”

As in the previous data, northern participants also invoke the myth of the nonaccent, both in explicit references to key figures such as news anchors, in (2), as well as implicitly, through the positioning of other accents as marked.

Another commonality is that both central and northern Ohioans overwhelmingly attend to southernness in language. The South was mentioned by 71 percent (n = 39) of northerners, not significantly different from the 80 percent (n = 52) of nonnortherners who mentioned it. Southern speech was frequently linked conceptually to the country, particularly in the southeast of Ohio as in (3), echoing the widespread ideology of the U.S. South as a locus of rural places, people, and identity (Preston 1997). Southeastern Ohio is an area geographically south within the state, but also the area most marked as socially Appalachian, being within the Appalachian region, lacking large cities, and being relatively poverty stricken.

Rural places were not limited to the southeast or even the southern parts of the state, however, as (4) and (5) demonstrate. Lima, mentioned in (4), is a relatively small city in the northwest area of the state.

Participants marked as country any areas outside cities, or even outside major cities. While some work has documented the use of a Country or Country Talk way of speaking as a positive, in-group marker (Hall-Lew & Stephens 2012), no self-identification was seen in these data. Even residents of rural areas who marked the speech of their town as country or otherwise distinct did so without including themselves in that category. Also unlike Hall-Lew and Stephens (2012), no distinction was made between country on the one hand and rural on the other. For example, Cincinnati, though often remarked on as distinct and sometimes marked as southern, was never described as country.

In contrast to the relatively consistent discussion of rural speech, the northern-based data differs somewhat from the rest in the treatment of urban speech. Among nonnortherners, the distinctness of urban speech was a repeated theme, with participants noting urban speech as divergent from an imagined norm, variously invoking youth, slang, and African American speech as explanations for this divergence (Campbell-Kibler 2012). Northerners show rates of commentary on urbanness (10 percent, n = 6) that are not significantly different from the nonnortherners (20 percent, n = 13). However, the character of these comments differed. Nonnorthern comments included both neutral comments and those invoking at least one of the themes of race, slang or youth. In contrast, the northern comments added to those two categories a third explicitly positioning urban speech in general as more standard, with each strategy used by two individual participants. Although the numbers of such comments are not sufficient to provide clear quantitative evidence, the existence of such comments in the north suggests that future work on the conceptualization of urban speech might be fruitful.

Cleveland and the Northeast

We discussed above the differing perceptions of central Ohio, a place that central Ohioans note more than northern Ohioans. Similarly, when nonnorthern Ohioans indicate the northeast, they tend to demarcate a relatively broad chunk, essentially lopping off the entire northeastern corner of the state as one region. In contrast, northerners, who were almost entirely from this area, showed more sensitivity to nuance, with some distinguishing Cleveland and its immediate suburbs from a broader northeast area, something not seen in the nonnorthern responses.

The differences in views of the north or northeast go beyond the relatively predictable observation that residents of an area observe more variability within it than do outsiders. One of the key motivations for our study was to better understand the popular understanding within Ohio of the dialectological boundary separating the Inland North and the Midland. This boundary is in linguistic terms a significant one, demarcating two varieties of American English that show a number of differences in vowel production and are believed to be moving further apart from one another (Labov, Ash & Boberg 2006; Labov 2010).

We analyzed the verbal responses by coding each participant into one of four possible categories—only three of which were evidenced—based on their description or lack of description of a boundary roughly comparable to that between the Inland North and Midland. The three categories found in the data were: no boundary or difference marked at all, a boundary between the two in which the north or northeast was verbally positioned as the region of note, and a boundary in which the central (or combined central-southern) region was treated as notable. A fourth logically possible category did not appear, in which both sides of the boundary were described as of equal note. It should be observed that, unlike the coding of the maps, regions mentioned only as a point of comparison for another region were not included, so that a comment like “X speaks differently than Y” was coded as marking only X as a region of note.

For simplicity, and due to the differential conceptual treatment of urban and rural speech, we judged any perceived boundary inserted between Columbus and Cleveland to represent such a division, regardless of its precise location or its extension through the northwest quadrant of Ohio. This classifies participants who marked a northwest region but no northeast region (five total: one from central Ohio, three from northeast Ohio, and one excluded from the current analysis for having lived in multiple regions) as not indicating such a division. This decision was based on the general emphasis on northeast Ohio as a locus of northern speech and the potential ambiguity of northwest-only regions between northern and rural, making classification more challenging.

The first category represents the simplest strategy for engaging with a border: to simply fail to construct it. Fifty-two percent (n = 34) of nonnortherners and 42 percent (n = 23) of northerners exhibited this response, proportions not significantly different from one another. These participants in some cases specifically noted the sameness between the two areas, as does the speaker in (6), with some hesitation. In (7), which represents one entire response to the map task, this sameness is not asserted, but the participant is coded as not building a distinction due to his lack of mention of Cleveland or the northeast. Both examples underline the situational dependence of these data and the at-times misleading clarity lent by categorical coding.

Another 27 percent (n = 15) of northerners join the rest of other Ohioans (48 percent, n = 32) in describing northern speech as worthy of comment, if not entirely accented. To the extent that linguistic detail was offered, most common were terms such as nasal (as in 8 and 9) and apparent references to the trap/bath vowel, i.e., a mention of Clevelanders’ “A’s” (as in 10 and 11). The latter were unusual, however, and most participants offered no linguistic detail.

Finally, the remaining 31 percent (n = 17) of northerners adopted a strategy not seen among other Ohioans at all: acknowledging a difference between northern speech and that which central Ohioans would regard as central (e.g., speech south of Mansfield), but marking the northern speech as normative and the central speech as divergent. This approach took one of two forms. Some speakers characterized speech in and near Columbus as different, usually in an inexplicable way, although note the reversal in (12) of Cleveland’s “fronter” speech to Columbusites’ speaking “a little farther back in their throat.”

A related strategy which also positioned the north as normative did so by simply understanding central Ohio as southern, as in (13). Note that Mansfield, proposed here as the northern boundary of the Southern accent, is in our study the demarcation between Inland North and Midland.

These results suggest that when northern Ohioans are prompted to consider regional variation within Ohio, a slim majority delineates a boundary which aligns broadly with the southern boundary of the dialect region(s) documented as North or Upper North (Carver 1987; Labov, Ash & Boberg 2006; Thomas 2010). Unlike central Ohioans, however, the northerners who do construct a boundary in this geographic area are divided between those who classify themselves as divergent, relative to the other side of that boundary, and those who classify themselves as normative.

Discussion

The results presented show that there is strong similarity overall in the ideological concerns of northern and nonnorthern Ohioans in terms of the sociolinguistic landscape of their state. They share a dominant north versus south orientation in terms of region, structuring their subregions of the state in terms of north and south or northeast, northwest, southeast and southwest. The predominantly central Ohio nonnortherners show a stronger sense of central Ohio as a region, while predominantly northeastern northerners show a more nuanced view of the northeast, marking out Cleveland and its suburbs as distinct from a larger northeast area. All Ohioans show a heavy concern for population density as an important sociolinguistic concept, particularly in marking rural areas as distinct, often stigmatized, and linked to the south. The two groups diverged somewhat in their conceptualization of urban spaces, with themes of young black and/or slang speakers present in both groups, but only northerners showing a rare competing discourse of urban centers generally as loci of standard speech.

The largest point of difference came in the conceptualization of the north, where northern Ohioans show a split, with some speakers positioning themselves as normative while others seeing their region as marked in contrast to an unmarked central Ohio. Central Ohioans, in contrast, are united in their perception of nonrural central Ohio as normative, only differing in whether they include the north as unmarked or note it as distinct. The restriction to the north of the perception of the north as normative suggests two possibilities that are not mutually exclusive. The first is that the decision of which speech is normative is derived from the triggering experience for the individual: direct experience or a link in a speech chain (Agha 2007). If they first understand an Inland North/Midland distinction as a result of direct linguistic experience, hearing and noticing some aspect of another’s speech, a speaker may be more likely to assume their own variety is normative. In contrast, a speaker whose first experience of difference is someone else remarking on the distinctiveness of their speech, or in some other metapragmatic move, may be more likely to accept that external assessment.

The second possible influence is the reaction of the individual as a stakeholder in a larger sociolinguistic system. A northern speaker will be impacted differently by a belief that their speech is nonnormative than by a belief that someone else’s speech is. With the data provided, we are not able to fully tease out these two possible influences, but the difference in personal stakes provides at least as plausible an explanation for the northern/nonnorthern difference as does a claim that there are systematic differences in the context of encounter for the two groups.

Given that explanation, these data offer some support for Agha’s (2007) point that speakers have highly interested stakes in the success and propagation of particular reflexive models, and that these stakes impact the ways in which models are circulated. In addition, however, they challenge the argument by Johnstone, Andrus and Danielson (2006) that “first order indexicals” in their system (linguistic features exhibiting patterns visible to analysts but lacking explicit or implicit awareness of difference) lack meaning entirely. Speakers who do not classify a form such as raised trap/bath as specifically Cleveland or northern do nonetheless have a social meaning for the form: the same meaning they associate with a lower, canonically Midland pronunciation, namely, normative and unaccented, as well as American and native English speaking.

The full characterization of that normative register will differ across communities or within them, depending on various factors. Some may see unremarkable forms as particularly educated, while others may see them as boring. The key point, though, is that the existence of a reflexive model that distinguishes Cleveland or northern speech from the norm does not give meaning to new forms from the ether. Rather, it threatens or enhances other form-meaning pairs, assigning new competing or compatible meanings, depending on the ideological relationships between the concepts involved. In the case of regional accent understandings in the United States, regionally-linked speech is in direct opposition to unmarked, educated, and standard speech. Regular users of Inland North forms encountering a model of the northern accent face not only the possible association of their speech with new meanings, but also the loss of existing meanings (note the similarity to notions of linguistic appropriation, as in Hill 1995).

The importance of this disruption is particularly visible in the current case, because the use of “unmarked” or “general American” speech forms has a rich social history, particularly in the Midwest. The use of the standard is associated with the usual baggage of education, articulateness, and correctness found in most understandings of standard language (Milroy 2000; Lippi-Green 2012). In addition, however, it invokes a host of social, racial, and political understandings regarding the racial purity of the Midwest (Bonfiglio 2002), tension between coastal and inland U.S. residents, and a view of the Midwest as real, authentic, or boring.

This larger framework adds weight to the social positioning of speech as accented or accent-free, suggesting that in this particular case, it may be that the added meaning is a relatively unimportant element, from the point of view of speakers. Few speakers are willing to label the north as accented explicitly, instead merely noting it as distinct, here or in Campbell-Kibler (2012). No insulting terms or characterological figures are offered, and the descriptions are diffuse. Unaccented speech, in contrast, is a central construct in U.S. standard language ideology, particularly in the Midwest. Put simply, the “stake” in the competing reflexive models for northern speakers may be less about sounding Cleveland and more about ceasing to sound like God and Walter Cronkite.

Judging which aspects are most important to actual speakers is beyond the scope of the current project, and indeed it is not clear whether it is possible to fully tease them apart in practice. Nonetheless, the comments that are available in the data provide a clearer picture of the positive valuation of unmarked speech than any negative valuation of the constructs of the Cleveland or northern accent. Yet the disparity between northern and other Ohioans in the assignment of markedness suggests that speakers have a stake in avoiding the notion of accent as applied to their own speech.

The meaning-bearing of unmarked forms is not limited, however, to the U.S. Midwest or to populations with notably high rates of linguistic security. We would argue that in general all linguistic forms carry some sort of social meaning, even those that do not excite attention via explicit commentary, distinctive metapragmatic framing, or the manipulation of forms in response to sociolinguistic pressures (for example, the shift from casual speech to reading a written passage). These less remarkable forms carry meanings that are unexciting or unexceptional in the majority of the contexts in which they are deployed. In other contexts, these meanings, always present, may become remarkable. For example, when in a rural area of another country, hearing speech very much like one’s own might prompt a traveler to surprise and strong positive (“Someone from home!”) or negative (“Tourists have invaded here too?”) feelings. Indeed, ideologies of where meanings like “normal” are remarkable or unremarkable can be quite revealing, as in the case of the public discussions of Barack Obama as notably “articulate,” which highlighted racist assumptions about Black speech as Obama’s race—rather than his education, position, or speech context—was taken to define his “expected” speech (Alim & Smitherman 2012).

Conclusions

The current study investigated the imagined dialectological landscapes of northern Ohioans, with the goal of better understanding their engagement with the construct of the northern or Cleveland accent recently documented as emerging in the understandings of other Ohioans (Campbell-Kibler 2012). The results show that while northerners share many ideologies about regional variation with their fellow Ohioans, they differ on their understanding of the Inland North/Midland boundary. While northerners were just as likely to construct a boundary resembling the Inland North/Midland boundary, half of those who did placed themselves as the imagined norm, constructing the speakers to the south of that boundary as accented or merely different. In contrast, all of the nonnorthern Ohioans who constructed a geographically similar boundary did so in the context of differentiating a northern way of speaking from a normative central way of speaking.

This illustrates the importance of individual speaker stakes in the reflexive models circulating in their corners of the social world. It also highlights the importance of normativeness or lack of accent as a valuable social meaning in and of itself. This is true both as a result of ideologies specifically about correctness and the standard, but as a general observation that unmarked or expected meanings nonetheless exist as links between language forms and social reactions (for example, lack of comment).