ABSTRACT OF THESIS PERCEPTUAL DIALECTOLOGY OF NEW ENGLAND: VIEWS FROM MAINE AND THE WEB Research into the dialects of the New England states Connecticut, New Hampshire, Maine, Massachuse
Sociocultural history of New England
While the broader history of European contact with the New World has been extensively documented, this project focuses on the English language’s development in the New World, especially in New England The use of English along the coasts of New England and across America can be traced back to the early 17th century, when ships’ crews landed ashore to provision themselves during trans-Atlantic fishing expeditions to Georges Bank The first British attempt to establish a colony in New England occurred in 1607 at Popham, Maine, but it failed due to climate challenges, notably the first winter It was not until 1620 that the British founded a longer-term settlement further south.
Massachusetts Bay Colony, which would develop into Boston and the surrounding area, stood as a crucial gateway for colonists arriving in New England English immigrants typically landed at this port before fanning out into the frontier, establishing a pattern of settlement that extended into the broader region Together with the Connecticut Colony at the mouth of the Connecticut River near modern-day New Haven, Massachusetts Bay Colony served as a cultural hearth—a starting point from which northern settlements dispersed Because of their central role in directing newcomers to America, these hearths, especially Massachusetts Bay, shaped the pace and direction of early colonial expansion.
Colony and Boston) have been cited as the source for New England culture; or, phrased
1 The Georges Bank runs from the coast of Cape Code up into the Canadian Maritimes, and plays an important role in the ecosystem of the area’s fisheries
4 another way: the culture of New England started at these hearths and then spread outward (Carver 1989, Steinitz 1989)
Wilbur Zelinsky, a cultural geographer, formalized the concept of cultural spread under the Doctrine of First Established Settlement, also known as the Founders’ Effect, as described by Stanford, Leddy-Cecere and Baclawski (2012) The Founders’ Effect posits that the first groups to settle a region imprint enduring patterns in language, religion, cuisine, and other cultural traits, shaping how later migrants diffuse and adopt local customs In this view, early settlement trajectories create lasting cultural landscapes, making the initial founders’ imprint a key driver of regional diffusion and cultural identity.
Whenever a new territory is settled or an earlier population is displaced by invaders, the defining traits of the first viable, self-perpetuating community shape the long-term social and cultural geography of the area, regardless of how small the initial band of settlers may have been Those pioneering characteristics—how the group organized itself, its economic base, and its cultural norms—set the template for subsequent settlement patterns, regional identity, and cultural landscapes that endure long after the first arrival.
Because the Massachusetts Bay Colony was the first successful and self-perpetuating English settlement, its culture became influential across New England The distinctive customs and institutions it fostered spread to the other New England states, shaping their growth and regional identity.
Zelinsky’s analysis of New England settlement shows that the process did not involve entering an empty territory but rather displacing an earlier population Even before formal colonization, sailors had contact with Native peoples, and as the English expanded into North America they repeatedly displaced earlier native groups (Woodard 2004; Vaughan 1965).
Evidence of sociocultural cohesiveness in this region is further explored in Kermes (2008), which shows how New Englanders collectively forged a regional identity that set them apart from Europeans and from their neighbors in the American South Kermes argues that the creation and ongoing maintenance of this regional identity was part of a broader process of forming a national American identity, yet her focus on a shared regional consciousness also relates to the way regional loyalties contribute to the larger national narrative.
Founders’ Effect and the present discussion According to Kermes, elements of the
2 It should be noted at this point that this project is following in the tradition of using the term Founders’
Effect as proposed by Zelinsky and used within cultural geography, which is consistent with the use of the
Within New England linguistic studies, the Doctrine of First Established Settlement is a central concept (Johnson 2010; Stanford, Leddy-Cecere and Baclawski 2012) The term Founders’ Effect can be easily confused with the Founders’ Principle, which describes genetic changes after a population bottleneck when a subgroup splits away from a larger population This biology-derived idea has entered linguistics independently (Mufwene 2008, 2001) As Mufwene (2001) notes, there is no substantial difference in how these terms are applied in linguistic contexts—the underlying principle remains the same: a group leaves the larger population and its linguistic variety propagates in a new environment.
The shared culture of New England’s early Puritan settlements centered on self-governance, education, and egalitarianism A key facet of this egalitarianism was language equality, a point Kermes notes in a summary of an article from an early Boston magazine This emphasis on linguistic parity helped bind the community and reflect the Puritan commitment to accessible knowledge and civic participation, shaping the region’s social and cultural fabric.
In contrast to Great Britain, where language divided the country along geographic and class lines, language in New England acted as a unifying force, with virtually all residents speaking Standard English The article argues that this shared linguistic standard created a sense of social cohesion, making speakers across the region sound as though they belonged to the English upper class.
Although the magazine's notes on actual language use cannot be independently verified, they reflect a contemporaneous view of the language New Englanders, particularly a segment of Bostonians, were perceived as sharing a single English variety—considered prestigious, effectively a "Standard" or upper-class form of the language.
Forming a distinct New England identity was not a smooth, disagreement-free process In fact, the settlement of the New England states reveals numerous disagreements that fractured the original colonies These tensions—driven by competing religious communities, economic aims, and political loyalties—helped shape a regional character that emerged through dispute as much as consensus.
Six modern New England states trace their origins to the Massachusetts Bay Colony Rhode Island was founded by Roger Williams after his exile from the Massachusetts Bay Colony over religious disputes (Bolton 1929) and has long been seen as a refuge for religious dissenters (Carver 1989) The Connecticut Colony also formed amid religious disagreements and ultimately broke away to become a prominent center in its own right (Carver 1989) Vermont, originally part of New France and ceded to Britain in 1763 (Bolton 1929), is the only New England state where the Founders’ Effect appears least applicable, though English settlers from Connecticut did push into areas of present-day Vermont along the Connecticut River into the Green Mountain region (Johnson 2010, Kurath 1939) New Hampshire gained independence after resolving a series of land disputes with the British government in the 1680s, which forced the Massachusetts Bay Colony to relinquish governance in the region (Adams 1921) Maine remained part of the original Massachusetts Bay Colony the longest among the modern states and experienced a lengthy period of secession attempts for various social and economic reasons before gaining statehood (Woodard 2004).
Masshole is one of six common pejorative demonyms for Massachusetts residents and is widely used across much of New England It points to a broader cultural distancing between the newer states in the region and the traditional New England hearth of Massachusetts The term's use is often seen as an assertion of a non-Massachusetts—and more specifically non-Boston—identity in the surrounding states, a point noted by Nagy.
By analyzing the Founders’ Effect alongside cultural diffusion across New England’s settlement and its later division into multiple states, we can trace how dialects respond to these sociohistorical forces If language is a cultural artifact that travels from hearths to new settlements, then New England offers a clear case for dialect development shaped by migration and regional partition Is there linguistic evidence—phonetic shifts, lexical innovations, and syntactic changes—that tracks this development from colonial times to distinct state varieties? Combining sociolinguistic data with regional history makes it possible to identify how diffusion and founder effects have left measurable traces in today’s New England dialects.
New England through a linguistic lens
New England's linguistic history stands as one of the most extensive in North American dialectology Dialect research began in the region in the early 20th century with Hans Kurath's Linguistic Atlas of New England (LANE), produced as part of the Linguistic Atlas of North America (Chambers and Trudgill 2008) Kurath's work focused on production data, published across several volumes and in the Handbook of the Linguistic Geography of New England (Kurath 1939), detailing lexical and phonological features of residents in all six New England states As is typical of atlas-based dialect studies, the data were collected and mapped to identify isoglosses and dialect regions, and researchers used knowledge of settlement patterns to interpret the results A fairly robust correlation emerged between the isoglosses and accounts of population movements.
Following this study nearly fifty years later, Carver revisited the isoglosses of New England through an analysis of data compiled in the Dictionary of American
Regional English (DARE), dating from the 1960s, shows that many lexical changes have occurred since the LANE survey, and several isoglosses once used to mark dialect boundaries are no longer evident Carver (1989: 27–30) attributes these discrepancies to an over-reliance on settlement patterns when defining the original dialect boundaries.
Of particular note, however, was the maintenance of the Eastern and Western New
Figure 2.1 shows the English dialect boundary By examining the two settlement hearths from which New England was colonized, the region appears relatively stable, and the Founders’ Effect can be seen shaping the dialect regions Settlements predominantly derived from the Massachusetts Bay Colony hearth form the eastern region, while western settlements tracing their lineage to the Connecticut Colony belong to the western region.
Figure 2.1 Generalized map of the New England dialects (Carver 1989: 31)
After Carver’s work, which addressed American speech broadly rather than focusing specifically on New England, research on New England dialects experienced another interruption It wasn’t until the early 2000s that a chapter in A Handbook of Varieties of English, by Nagy and Roberts, documented the phonological features of New Englanders, marking a renewed scholarly interest in the region’s distinctive speech patterns.
The most recent large-scale study of dialectal variation in New England was conducted as part of the TELSUR project, which produced The Phonological Atlas of North American English In this work, the traditional East–West divisions of New England were maintained, and Carver’s identification of a more peripheral Northern New England region was preserved (Labov, Ash, and Boberg 2006; see Figure 2.2).
A common thread to all of the aforementioned studies is a shared finding that dialect boundaries within New England have remained stable since the publication of
LANE As a means of explaining this consistency, dialectologists have often invoked the Founders’ Effect (Stanford, Leddy-Cecere and Baclawski 2012; Labov, Ash, and Boberg
Figure 2.2 Regional dialect map showing the division of Western and Eastern New
Findings on New England dialects reported by Labov, Ash, and Boberg (2006) have begun to face critique: not all field-workers in Kurath’s initial study were equally skilled at collecting and classifying data, leading to some inaccurately categorized regions in southern New England (Johnson 2010) Moreover, mounting evidence challenges the Founders’ Effect as a universal explanation, with sections of New England—most notably New Hampshire—shifting away from the Eastern New England dialect While the changes are not large-scale, these studies document local dialect shifts, showing areas in New Hampshire, once grouped with Eastern New England alongside much of Massachusetts, increasingly aligning with Western New England features (Stanford, Leddy-Cecere and Baclawski 2012).
Examples of this shift are most evident in research on the dialects of New Hampshire When comparing differences in the production of variants between Boston and New Hampshire, Nagy concludes that some of the differences result from establishing a separate identity that is distinct from the neighboring urban center (Nagy, 2001).
A study by Stanford, Leddy-Cecere, and Baclawski, surveying dialectal features along the Vermont–New Hampshire border, reports that "for younger speakers the most salient east-west contrasts are rapidly dissolving" (2012:157) The result calls into question whether the traditional dialect boundaries in the region remain intact However, these studies have not addressed how speakers themselves perceive these borders.
Research on dialect variation in New England has not been entirely neglected; instances exist where New Englanders demonstrate knowledge of dialect differences that appear in production studies (Stanford, Leddy-Cecere, and Baclawski 2012) Yet studies focusing specifically on perceived dialect regions in New England remain rare, with only one such study identified to date examining how dialect regions are perceived by and of New Englanders, and limited to speakers in New Hampshire and Maine (Ravindranath and Fernandes 2014).
In their study of New England regional dialect perceptions, Ravindranath and Fernandes (2014) employed the tools of perceptual dialectology to elicit the regions where residents of New Hampshire and Maine identified dialectal variants In addition to gathering geographic data on where these regions existed in the minds of the respondents, attitudinal data was collected showing how each dialect region was perceived Regional labels including “hicks” and “Massholes” were considered for how the perceived speech was regarded, and social factors of the respondents were also brought into the analysis One of the findings from this analysis was the existence of variation in the perceptual regions based on age, as seen in Figure 2.3 (showing ratings for correctness, with the left most image representing a composite map for the oldest respondents, the right most the youngest respondents and warmer colors representing more positive ratings) This result raises questions as to whether variance in the location of perceived regions can be attributed to generational shifts reflecting language change in progress or changes in the interactions of people across these regions This study shows how information about dialect regions can be elicited from speakers and used to further inform linguistic
3 Hartley (2005) is a PD study of residents of Boston, but does not (1) encompass New Englanders as a group and (2) does not specifically look at New England in its design, instead focusing on the nation
This study mirrors the approach described for the project in this thesis, illustrating how similar research can proceed A solid understanding of the theoretical practices of perceptual dialectology is required before undertaking such work, because these concepts shape the research design, data collection, and analysis, anchoring the findings in the field’s theoretical framework.
Figure 2.3 Heat maps showing views of linguistic “correctness” across three age categories (Ravindranath and Fernandes 2014)
Theoretical Framework
Perceptual dialectology (PD) extends the folk linguistics tradition by drawing on non-linguist respondents’ knowledge to gauge language from a lay perspective Although to theoretical linguists this may seem trivial or futile since lay opinions are not scientifically grounded, proponents argue that understanding the social aspects of human language requires attention to how structural variation is perceived and evaluated, just as it is to measure these features as they manifest in speech Integrating social recognition of language into models of language variation can yield more robust theories about how variation exists and changes over time (Niedzielski and Preston 2003).
The specific goal of perceptual dialectology within this tradition is to understand how speakers view the spatial distribution of dialects (Preston 1989) This approach reverses traditional dialectology: instead of the linguist drawing isoglosses, the researcher asks speakers how they would delineate the boundaries Principles for these methods derive from the mental maps used in cultural geography (Gould and White 1986), where individuals are asked to produce maps of specific environments to reveal how people store and access spatial information.
These maps offer insights into how people perceive the locations of goods and services and which landmarks they recognize, revealing patterns in regional awareness They have often been used for humorous contrast, such as how residents of America’s largest city imagine the rest of the United States, a notion popularized by the New Yorker’s 1976 map showing how New Yorkers view the country’s geography Such maps have been produced by researchers and media outlets to explore public perception of place and space.
Daniel Wallingford created a map that distorts the North American continent to reflect a Bostonian’s view of the United States (Figure 2.5) This example links naturally to sociolinguistics when viewed through Labov’s concepts of indicators, markers, and stereotypes in linguistic variation In perceptual dialectology (PD), the linguist probes what markers and stereotypes people associate with different dialects and regions as perceived by non-linguists.
Figure 2.4 Cover of the March 1976 New Yorker (Mappery)
To study regional linguistic variation, participants map dialect regions by drawing or circling areas on a blank map and labeling where they perceive different dialects to occur They then describe the labels they assigned to each region, and, in many studies, rate the language varieties on attributes such as pleasantness and perceived correctness relative to state boundaries (i.e., political borders).
“How would you rate the English spoken in Idaho in terms of pleasantness?”; Preston
Wallingford also produced a map offering a view of the United States from New York, using a visualization similar to the earlier map, and both maps were created prior to the publication of the New Yorker cover page (Preston).
1989) While rating of states was the foundation for the attitudinal aspect of understanding regional differences, there has been a push by some perceptual
Figure 2.5 Daniel Wallingford’s map of the view of the United States from Boston
George Glazer Gallery suggests dialectologists should instead collect ratings for the actual regions drawn by respondents (Cramer forthcoming) In either event, this method enables the dialectologist to capture mental snapshots of where language varieties are perceived on the national scale (Preston).
Regional and state-level studies consistently show that certain dialect regions recur across respondents, with the American South especially prominent in national-scale analyses (Preston 1989; Evans 2011; Cramer 2010) Even when respondent locations vary, these regions reappear, so that people from places like Michigan and Hawaii share a sense of where "the South" is, even if its exact position on their mental maps differs (Preston 1989) This replicability has yielded a nation-level template for United States dialect regions that can be used for cross-study comparisons (Figure 2.6) Once the maps are gathered, they can be compared with extant maps to reveal how other factors influence perceptions of these locales, offering deeper insights into regional identities and linguistic landscapes (Montgomery).
(2007) demonstrated that perceptual dialect maps in England closely followed traditional North-South distinctions, mirroring divisions found on other maps demonstrating social
13 differences Cramer (2010) has shown that the major regions identified in the
Commonwealth of Kentucky also reflect larger discourses of cultural regions in that state
Figure 2.6 Template of perceptual regions in the United States (Preston 1989: 127)
Perceptual dialectology begins by producing an individual map, but the real insight comes from comparing and aggregating multiple maps While isolated maps have value on their own, GIS-enabled aggregation lets researchers digitize and georeference the maps, import them into a GIS, and code regions within a common geocoordinate system for rigorous comparison Overlap analyses then reveal shared perceived regions across respondents, such as combining areas labeled “the South” to form a region that reflects the level of agreement on its placement This GIS-based approach poses two challenges for the dialectologist: first, matching regions across maps when similarly named areas may actually denote different locations; second, reconciling naming variations with spatial boundaries to support meaningful cross-map comparisons.
Within the same area, the label "twang" is applied to denote a particular regional speech feature The decision to use this label is largely left to the dialectologist, based on their best judgments informed by regional research and other descriptive data collected from respondents (Evans 2011, Cramer).
Digitization poses a second major challenge—the laborious process of tracing scanned images into analysis software Acknowledging this hurdle, Preston and Howe (1987) conducted a study with an early touch-screen interface that let respondents start with a digitized map and have it directly imported into an analysis program Although this approach was abandoned at the time due to technological constraints, recent advances—especially the widespread use of touch screens in everyday devices—invite a reconsideration of digital applications Yet to date, no studies or PD-specific tools have surfaced.
Beyond merely speeding up map digitization, PD practitioners have strong incentives to adopt new tools GIS wasn’t built with dialectology in mind, yet dialectologists have repurposed it to amplify their work, turning spatial analysis, data integration, and visualization into core research capabilities Its contributions to the field’s methods have been immense, driving methodological rigor and new insights into dialect variation.
Perceptual dialectology has not been fully served by GIS yet First, GIS does not readily allow respondents to draw on maps as in Preston and Howe (1987) and, although ArcGIS and QGIS have recently become touch‑screen/tablet accessible, interactive map‑drawing remains limited Second, deploying GIS in large‑scale research is still difficult, which is a concern for dialectologists who want to survey broader regions beyond their home institutions The desire to crowd‑source data is not new to the field; dialectologists have been interested in crowd‑sourced data since 1876 and Georg Wenker’s atlas of regional German dialects (Chambers and Trudgill 1980) While crowd‑sourcing programs such as OpenStreetMap exist, they are not designed for perceptual dialectology, making PD an active area of growth and research.
Finally, the contributions of PD to the field of linguistics require mention Such research has given insight into how individuals view linguistic variation as a means of
15 constructing identity and associating with different regions (Cramer forthcoming, Cramer
The salience of dialect regions to speakers and their role in constructing identity often yields a fluid picture, as shown by Cramer’s forthcoming work, which indicates that both out-group and in-group identities can be read through the placement and labeling of perceived dialect regions Montgomery (2012) further demonstrates that regional borders shift depending on a respondent’s proximity to borders that are more or less culturally shared, and that this closeness also influences how dialect regions are labeled—a finding supported by geographic research on spatial perception and evaluation (Gould and White 1986) PD has aided in identifying new dialect regions, as illustrated by the case of emerging dialect regions.
Manchester dialect first identified through perceptual maps elicited in Montgomery
(2007), a change-in-progress that could easily have been overlooked by linguists
This section outlines the methodological approach used to elicit perceptual maps of New England's dialect regions, blending established techniques from perceptual dialectology (PD) with procedures developed specifically for this thesis The PD-derived methods employ projective elicitation tasks and participant-generated perceptual data to chart how speakers perceive regional speech variation, while the thesis adds unique elements such as tailored stimulus materials, iterative validation, and integration with demographic and geographic information to improve accuracy and spatial resolution Data were collected through structured interviews and surveys, followed by quantitative aggregation and GIS-based spatial analysis to delineate perceptual boundaries among dialect regions The resulting perceptual maps illuminate the lay perception of dialect differences across New England and provide a robust, data-driven representation of regional speech variation.
Online data collection
We conducted the first experiment using an innovative online approach to collecting perceptual maps, enabling data collection from a larger, geographically diverse sample and eliminating the need for a secondary digitization step to input maps into GIS The survey was built with Qualtrics using available templates, and invitations were distributed via email to colleagues at other academic institutions as well as through social platforms like Facebook and Reddit A key advantage of Qualtrics for this study is its automatic reporting of IP addresses for unique visitors, allowing researchers to filter out responses from users outside the target area Consequently, any responses originating from IP addresses not associated with a New England state could be removed from the final dataset to minimize geographic noise and ensure the results reflect the intended region.
The survey consisted of three sections: first, an instructional video (link in Appendix A) showing how to use the software to perform the mapping task; respondents were then asked to create a map based on these instructions and to provide information about the regions drawn on the map; and finally, the survey collected attitudinal information through questions described below.
The first section included a link to a website called ImageBot (FlamingText) ImageBot was selected for the experiment because it allows users (in this case, the
In the study, 17 respondents uploaded an image, used a web-based interface to annotate or draw on the image, and exported the edited image back to their computers; the ImageBot interface is shown in Figure 3.1 Participants also downloaded a map of the six New England states from a Dropbox-hosted link, with the map produced by the Massachusetts Office of Geographic Information and presented in Figure 3.2 (MassGIS 2008).
Respondents were asked to draw areas on a map representing distinct speech varieties they perceived, labeling each area with descriptive names that reflect its type of speech They then saved the map to their local computer and uploaded the result to the survey Next, participants listed the names of the regions they had identified and briefly described their perceptions of the speech in each region in an open-text box The survey template used for this process is provided in Appendix B.
Figure 3.1 The ImageBot interface shows that respondents were asked to provide a demonym for residents of each New England state and to rate the speech of each state on qualities such as intelligence, pleasantness, friendliness, correctness, and similarity to their own speech, following the tradition of perceptual dialectology studies (Preston 1989) Responses for the similarity quality were collected using a seven-point Likert scale with the labels “Not at all like me,” “Not like me,” “Not much like me,” “Neutral,” “Somewhat like me,” “Like me,” and “Just like me,” while the other qualities were measured with a sliding bar.
Figure 3.2 Map presented to respondents (MassGIS 2008), shading of interior not present in original they could manipulate with their mouse between the extremes of “Very [the quality]” and
Data quality was very poor An example of the sliding input is shown in Figure 3.3 The choice to have respondents rate the states rather than the regions reflected the established tradition of rating states in PD research that had been identified earlier.
Recent research, including Cramer’s forthcoming Kentucky study, has asked respondents to rate the regions they themselves produced This approach was adopted in the second experiment, but it was not used in the current iteration due to software limitations that prevented feeding the drawn-region data back into the survey without risking data loss in follow-up surveys The survey concluded with an open-ended text box, allowing respondents to share any comments they wished about the experiment.
Figure 3.3 Sample of sliding scale input
Traditional pen-and-paper task
The second, follow-up experiment was conducted as fieldwork in the summer of
In southern Maine in 2015, respondents were recruited through the researcher's friends For the draw-a-map portion, a printout of the online experiment map was used, and participants were asked to provide the same content regarding perceived speech regions Each region identified was followed by an individual sheet of paper containing a brief questionnaire Attitudinal data on perceived similarity and several evaluative dimensions—correctness, pleasantness, friendliness, and intelligence of the dialect—were collected using the same Likert scale A sliding scale was not used because Qualtrics automatically assigns numeric values to slider positions in the online version, whereas the pen-and-paper version lacks this feature Two open-ended questions completed the questionnaire: the first asked what factors participants considered when labeling a region, targeting mental representations and stereotypes of the dialect regions, and the second invited any further comments about the region.
20 designed to allow for more open-ended commentary by the respondent An example of this questionnaire is contained in Appendix B
Maps collected by this method were scanned and digitized, and each Likert item was converted into a numeric score on a 1–7 scale, with 1 indicating the lowest extreme ('Very uneducated') and 7 indicating the highest ('Very pleasant') The resulting numeric data were then used to compute the results described in Sections 4 and 5.
Data processing
Across both versions of the experiment, respondents’ digital maps (or their drawings) were uploaded to ArcGIS and georeferenced to the coordinates of the original map produced by the Massachusetts Office of Geographic Information, enabling geospatial analysis of the regions within the program This georeferencing aligns the abstract lines drawn by respondents with the original map coordinates, ensuring consistency across images Because the study had a small number of respondents (an issue addressed in Section 5), all regions drawn by participants in both data collection methods were included in the subsequent analysis Attitudinal data from the Likert scales were tabulated, with basic statistics such as the mean and mode calculated, since the limited sample size precludes more robust statistical analysis.
The results of the two studies are summarized in the following two subsections, and a direct comparison of these results, along with a deeper examination of the initial findings, will be presented in Section 5.
Online Experiment
In the online version of the experiment, 14 usable responses (n=14) were obtained from 317 unique IP addresses, yielding a response rate just over 4%—below the project’s initial target discussed in Section 5 Data quality issues led to the removal of seven surveys that did not produce mental maps (labels with no boundaries or color swatches not corresponding to described regions) and five surveys from IP addresses outside New England Respondents ranged in age from 18 to 52, with 5 identifying as female, 8 as male, and 1 as other At the time of the survey, six respondents were in Massachusetts, two in New Hampshire, three in Maine, two in Rhode Island, and one in Connecticut Table 4.1 provides a summary of these results.
Table 4.1 Summary of demographic information, first experiment
State of Respondent No of Respondents Gender No of Respondents
In the first section of the survey, the draw-a-map task yielded a total of 74 regions across all maps The minimum number of regions on a map was two (observed in two respondents), while the maximum was nine (found in a single map) The average number of regions per map was just over five, and the mode was five regions Figure 4.1 provides a composite view of all boundaries drawn to represent these regions.
Figure 4.1 shows a composite of all lines drawn for the online experiment Because respondents were asked to label and describe the regions in an open-ended way, it is challenging to consistently connect the same regions across different maps This ambiguity complicates cross-map region alignment and highlights the need for standardized labeling when comparing regions across maps in online studies.
Across the data, respondents either refrained from labeling regions explicitly and instead described the dialect they had in mind, or they used explicit labels but offered little detail about how they delineated those boundaries Some regions appeared on only a single map and did not clearly correspond to similar regions on other maps, making direct cross-map comparisons difficult unless the analysis relied on the researcher’s intuition Despite this, it remains valuable to understand which regions were most salient as perceived speech areas during the task To this end, labels and descriptions were reviewed for key terms repeatedly used to designate perceived isoglosses; while many terms appeared only once or twice—reflecting varying levels of distinction in each respondent’s perceptual space—certain terms appeared more frequently across maps.
Across the dataset, Western New England and a Connecticut–Rhode Island mix appear three times, the designation of a "Standard" or "Generic" region occurs four times, Maine by itself appears five times, and Boston functions as a primary descriptor eight times, with these patterns summarized in Table 4.2 When modifiers are introduced, other regions gain prominence—Vermont receives three different designations, Maine is divided into North/South and Down East divisions, and New York City appears as a descriptor for Connecticut and Rhode Island Several areas show some influence from Boston, a trend visible in the respondent labels in Figure 4.2 An additional feature on the maps is the appearance of a "French Canadian Influenced" region; although not mentioned on every map, two maps note French influence along the United States–Quebec border. -**Support Pollinations.AI:**🌸 **Ad** 🌸 Discover [Pollinations.AI](https://pollinations.ai/redirect/kofi) for smarter content rewriting and SEO optimization—tailored for regional analysis articles like yours!
Table 4.2 Most frequent region labels, first experiment
Figure 4.2 Map drawn by a 25 year-old male from Massachusetts
Looking at individual maps reveals several interesting patterns in how Vermont’s dialect is represented Figure 4.3 shows a respondent who uses the Vermont state lines to emphasize its distinct dialect features, setting Vermont apart from the rest of New England Another participant, in Figure 4.4, isolates Vermont as an area of unfamiliarity—worth mentioning on the map even though no specific dialect features are identified By contrast, Figure 4.5 presents a respondent who groups Vermont with New Hampshire, suggesting a shared dialect that may reflect a “standard” version of English.
25 Figure 4.3 Map drawn by a 19 year-old male from Massachusetts
Figure 4.4 Map drawn by a 52 year-old male from New Hampshire
Figure 4.5 Map drawn by a 23 year-old male from Connecticut
Note that respondents used ImageBot tools differently: although the instructions asked participants to use the draw feature, some explored alternative options to complete the experiment Figure 4.6 depicts a respondent who employed the auto-shape features to outline their regions, while Figure 4.7 shows another participant using color-coding to identify regions These maps were included in the subsequent analysis because they still displayed dialect regions, even though their creation involved slightly different methods; this variation will be discussed further in Section 5.
Demonym data from the second section of the survey revealed a mix of predictable and unique terms Many responses used standard labels, with “New Englander” for someone from New England and “Rhode Islander” for a Rhode Island native Other respondents offered more unusual names: for instance, one respondent used “Yankee” to refer to New Englanders in general, while another restricted the term to residents of Vermont In some cases, respondents did not provide a demonym.
27 Figure 4.6 Map drawn by a 23 year-old from Massachusetts
Figure 4.7 Map drawn by an 18 year-old female from Massachusetts
In the study, respondents suggested terms for residents of each state; while some offered non-derogatory descriptors, others used insults such as “Hick” or “Yankees fan.” A particularly common term was “Masshole,” the demonym for Massachusetts residents, which appeared in ten responses, including two of the six respondents from Massachusetts.
The final portion of the survey, which rated similarity, pleasantness, friendliness, correctness, and intelligence, required special attention after it was discovered that the friendliness item duplicated the pleasantness item due to an import error from an earlier trial version; as a result, all instances of the friendliness item were excluded from analysis 6 Only nine of the fourteen respondents completed this section, with six from Massachusetts, two from New Hampshire, and one from Connecticut, creating a very small and unbalanced sample that prevents any robust inference or statistical validation While it might be tempting to discard these results, they still provide useful insight for future research For example, Massachusetts posted the highest mean score for similarity, but Connecticut led in pleasantness while Massachusetts had the lowest; the Massachusetts responses skewed the results, with two respondents rating the state at the extreme of unpleasant and the rest near neutral, and no positive ratings.
Traditional experiment
For the traditional version of the experiment conducted through fieldwork, thirteen respondents in Maine (n) were interviewed and asked to complete the task
Of these, eight identified as male, four as female, and one left the question blank Ages
5 The assumption that this is intended as a derogatory term is largely influenced by the author’s own experience growing up in the Bath/Brunswick region of Maine
6 The scores for both questions were remarkably similar in terms of raw values, showing that respondents were indeed answering the same question
Participants ranged in age from 29 to 66 Five respondents were based in Portland, Maine at the time of the survey, while eight were located in the Bath/Brunswick area of Maine’s Midcoast region The demographic data are summarized in Table 4.3.
Table 4.3 Summary of demographic information, second experiment
Region of Maine No of Respondents Gender No of Respondents
Across all maps in the draw-a-map portion of the survey, 34 regions were identified, averaging just under three regions per map The minimum number of regions on any map was one (appearing on three maps), while the maximum was six (on one map) The most common result was two regions, identified on four maps For the maps with only one region, one identified Massachusetts as the sole prominent dialect area, another highlighted a portion of northern Maine, and a third described a larger area that grouped several states together.
“All sound[ing] the same” (Figure 4.8) Figure 4.9 provides an overview of all boundaries that were drawn in this task
Figure 4.8 Map drawn by a 33 year-old in the Bath/Brunswick region
7 The Midcoast region begins (roughly) in the Bath/Brunswick area and runs along the coast to slightly east of Rockland
Although the labeling procedure for regional analysis followed the same guidelines as the online task, the map labels were easier to compare because they lacked narrative text; however, some labels remained difficult to align with others, such as "Lobsters taught us English" and "We enunciated our 'u's as in human" (Figure 4.10), whose connections to labels produced elsewhere are not clear The region descriptions provided by respondents offer little extra insight into how these areas relate to those identified by others, and in particular the latter label was described as being selected based on negative reception of the people and their speech, a rationale that does not readily map to the regions chosen by other respondents Furthermore, there is no additional information suggesting linguistic features that could link this perception to a specific dialect region Therefore, it is difficult to determine which linguistic—and not necessarily cultural—factors created an association with a dialect region for this respondent.
Figure 4.9 Composite map of all lines drawn for the traditional experiment
Figure 4.10 Map drawn by a 30 year-old male in the Bath/Brunswick region
Analyzing labels and descriptions to identify salient regional dialects, the study found Boston as the most frequent label, appearing on eight maps, with an “Outer-Boston Talk Region” on a single map (Figure 4.11) Down East (Downeast) followed, appearing on four maps, while French or French-American labels appeared on three maps Other labels were rare, with New York and Aroostook County (or “The County”) appearing on two maps each These patterns are summarized in Table 4.4 Across respondents, personal experience or media exposure was identified as the primary factor shaping which region they selected.
8 The Down East region is also a prominent region in terms of local discourse, including a popular magazine named after the region: Down East Magazine
Table 4.4 Most frequent region labels, second experiment
The map drawn by a 30-year-old man from the Bath/Brunswick region highlights several noteworthy patterns, with non-rhoticity emerging as a key factor in identifying dialect regions This theme recurs across multiple maps, including Figure 4.8, and is echoed by respondents who emphasize non-rhotic features in their own data visualizations For example, the author of Figure 4.12 explicitly drew attention to the presence or absence of “r” sounds in the phrase “car in the yard,” a nod to the oft-cited Boston pronunciation “Pahk yah cah in Havahd Yahd.” Two other respondents likewise spotlight non-rhotic features when generating their maps, underscoring the central role of this feature in delineating regional speech patterns.
Figures 4.13 and 4.14 show that 33 maps reveal how respondents used non-rhoticity to identify distinct Maine regions, with the 'Ah' variant marking the pronunciation shift where 'er' is pronounced as 'ah', and none of the maps pertain to the Boston region.
Figure 4.12 Map drawn by a 48 year-old male in the Bath/Brunswick area
Figure 4.13 Map drawn by a 37 year-old female in the Bath/Brunswick area
Figure 4.14 Map drawn by a 29 year-old male in the Bath/Brunswick region
Results for the region-rating task were derived from questionnaire responses As in the previous experiment, the small sample size limits meaningful statistical inferences, and the variability is amplified because not all respondents rated the same regions Consequently, this section focuses on the two most frequently named regions: Boston and Down East Of the eight respondents who identified Boston, the scores for similarity and other qualitative features were slightly below neutral, with mean values around 3 (where 3 reflects a neutral/correct rating).
Pleasantness received a score of 3.75, a position between somewhat unpleasant and neutral on the rating scale The four respondents who identified with Down East showed much more variable results Across friendliness, pleasantness, and perceived similarity, the Down East region was rated between neutral and only mildly positive for these features.
Friendliness was the highest-rated factor, with a mean of 5.25, slightly above "Somewhat friendly." Similarity and pleasantness followed with mean scores of 4.5 and 4.25, respectively, just above "Neutral." In contrast, the region was poorly perceived in terms of correctness and education, signaling weak overall perceptions in those areas.
35 region had a mean value of 2.5 (a value between “Somewhat incorrect” and “Incorrect”) while the level of education had a mean value of 2.75 (in this case being between
Respondents appear to hold a generally mild negative view of Boston speech, yet they react more strongly to a local dialect when judging it, a distinction captured by the terms "Somewhat uneducated" and "Uneducated." Table 4.5 is provided to visualize these results, showing that lower values correspond to more negative ratings, with a value of 4 marking a point on the scale used in this assessment.
“Neutral.” This perceptual data will be further explored in Section 5
Table 4.5 Summary of attitudinal data for Boston and Down East regions
Feature Mean for Boston Mean for Down East
Summary
Across both experiments, although the response pool was modest, Boston consistently stood out as a prominent dialect region—more prominent than any other area regardless of how the experiment was delivered The attitudinal data, however, did not rate Boston positively: when evaluated as a standalone region in the second experiment, reception was tepid, and when considered within its larger geographic placement in Massachusetts in the first experiment, it again drew negative marks Given the design differences between the two experiments, a direct comparison is limited, but the following section analyzes the data and highlights these patterns.
Before launching into the full data analysis in Section 4, a methodological issue must be addressed: the online version of the experiment produced a markedly low response rate Of the original 317 responses, 291 were excluded because, after consenting to participate, they provided no further information This shortcoming stems from the survey design rather than the research method Several respondents reported technical problems using ImageBot on their own devices; although the pilot run detected no errors, Reddit comments about the survey link referenced an inability to use the image editor Because ImageBot is maintained by a third party, remedies were limited, and we lacked information about users’ operating systems and browsers Additionally, the survey was perceived as cumbersome by many potential participants—Reddit feedback described the video as “tedious,” and several comments noted the substantial time and effort required—leading to high dropout and lower engagement The outcome is a reduced response rate with potential self-selection effects, and the Reddit comments cited are provided in Appendix A.
Online deployment faced a notable challenge in users’ perceptions of the study’s legitimacy While some potential respondents questioned the value of this kind of perceptual research, others reacted negatively to a specific factor: the researcher’s affiliation with an institution outside the local area Upon clicking the link, respondents encountered a disclaimer and consent page that clearly stated the study was being conducted by a linguist affiliated with an external institution These legitimacy concerns can impact participation and data quality, underscoring the need for transparent disclosures and local relevance in future online surveys.
University of Kentucky This prompted one commenter to associate problems with the survey with their negative perceptions of Kentucky 9 (a state well outside of New
England) while another commenter assumed that explaining that “most” New Englanders spoke Standard American English (providing a link to an external website to support their
9 Kentucky is also commonly associated with the American South dialect region, a region that regularly receives negative appraisals in PD research (c.f Preston 1989)
Although some issues can arise in a face-to-face survey, they were strikingly more evident in the online survey This pattern may reflect the perception of an outsider researching the language of another linguistic group, suggesting that the outsider is seen as part of a different linguistic community Unfortunately, this challenge is not easily mitigated in a virtual environment.
Although the first experiment faced delivery problems, its results do not indicate that future online perceptual dialect (PD) experiments should be discontinued, as the findings were not markedly different from the second experiment in eliciting perceptual maps of dialect regions; in fact, the online version produced the highest average number of regions and the most regions on a single map Across access patterns, respondents who could reach ImageBot's website were able to generate perceptual maps, and the maps produced online were just as detailed as those drawn on paper, yielding similar results in terms of prominent regions and descriptors.
Across experiment types, the status of Connecticut, Rhode Island, and Vermont stood out as a point of reference Some respondents stated no familiarity with all three states (Figure 4.4), while others largely omitted them (Figure 4.13) Connecticut and Rhode Island were often separated from the rest of New England by the perceptual boundaries drawn by respondents, and these two states (notably Connecticut) were associated with a non-New England region, New York, by both online and in-person participants, suggesting they do not share the regional speech of the other New England states Vermont is notable in that, while one respondent placed Vermont with New Hampshire in the same dialect region (Figure 4.6), other responses did not consistently align Vermont with that region.
10 Interestingly, this may also reflect a perpetuation of the belief held centuries ago that all of New England spoke the same, standard variety of English as discussed in Section 2
In a sample of 38 participants that included Vermont, Vermont stood out as distinct from its geographic neighbors Although Stanford, Leddy-Cecere and Baclawski (2012) suggested that the distinctions between Vermont and New Hampshire dialects (Western and Eastern New England, respectively) were dissolving, these experiments' results indicate that such dissolving is not reflected in the regional perceptions observed here.
The status of these three states raises questions about central and peripheral New England dialects and how these states are linguistically related to their neighbors When respondents were asked to label the dialects of New England, these regions remained largely uncharted because there were insufficient responses—or in some cases, no responses—from residents of these states, which prevents examining how they view their relationship to the rest of the region.
Boston was the most named region across perceptual maps, whether data were gathered online or on paper, yet its location and influence are not uniform across respondents Figure 5.1 shows an overlay of all online-collected maps labeling Boston, while Figure 5.2 presents the corresponding overlay for maps collected in person A black star, added to each map to mark the city’s location, does not appear in the original maps Darker shading indicates a higher level of agreement among respondents about Boston’s prominence.
Figure 5.1 Overlay map of the Boston region, as collected from online respondents
Figure 5.2 Overlay map of the Boston region, as collected traditionally
Two trends emerge when these maps are compared: one shows the Boston dialect spreading from the city along a north–south coastal belt, while the other places the dialect’s core in Boston with an east–west reach across Massachusetts Rather than a simple online-versus-in-person contrast, these patterns reflect geographic location: Maine residents proposed an in-person view in which the Boston dialect does not extend into Maine, whereas online respondents—largely based in Massachusetts—placed the region along the coastal corridor of Massachusetts, New Hampshire, and Maine This suggests that Massachusetts sources view the Boston dialect as extending beyond the state, while Mainers see their own linguistic varieties as distinct from Boston’s, perhaps distancing themselves from the urban center as suggested for New Hampshire by Nagy (2001) How Mainers perceive their variety as distinct remains unclear.
Distinguishing Maine varieties from the Boston variant is complicated by how regions are described through non-rhotic features Some respondents group several New England states and describe them via non-rhoticity (Figure 4.8), while others emphasize that these regions have this feature but are not connected (see Figures 4.13 and 4.14) Unfortunately, it is not clear what
40 other factors cause these regions to be distinct to the respondents as such information was not explicitly stated However, it is possible that this may be a result of proximity For the respondents in Maine, the non-rhotic regions may be more salient for their non-rhoticity due to their existing within the same state, while Boston is two states away
Following the idea that proximity may have some influence on perceptions
(Montgomery 2012), taking a brief aside and considering a finding particular to the traditional version of the experiment gives an interesting view of a regional dialect of Maine Down East, which was the second most labelled region for those that hand-drew their maps (Table 4.4, but it was also mentioned in the online maps as well), appears to cover a very wide area with only sections of overlap In Figure 5.3, which shows an overlay of all regions labelled Down East in the second experiment, we can see that there is a fair amount of agreement between respondents in that the region is coastal and eastern in the state However, the extent to which that region expands beyond the eastern- most corner of the state shows some variation Some respondents suggest that it is constrained to only the most eastern portion of the state (representing roughly an area of Hancock, Penobscot, and Washington counties), others suggest that it expands down through Maine’s Midcoast region and one respondent identified it as being present along the border with New Brunswick, Canada and up into the inland region of Fort Kent, Maine 11 It is unfortunate, then, that the sample size of this experiment was as small as it was, as it would be interesting to see just how extensive the region of Down East is perceived to be by Mainers
Given the data presented in this and the previous sections, what can we say about the perceptual landscapes of dialects for New Englanders, and what courses are open for future research into the region? The next section addresses this question
The precise Fort Kent location on the map isn't essential for our purposes What matters is that Fort Kent lies in Maine's far northern and inland region, well away from the Down East area traditionally associated with the term.
41 Figure 5.3 Overlay map of the region labelled Down East in the traditional experiment
6 Conclusion and Directions for Future Research