How Canadian was eh? A baseline investigation of usage and ideology

1. Introduction

Eh is a confirmational – a multifunctional pragmatic marker that typically functions to seek addressee confirmation of the truth (or knowledge of the truth) of the proposition that has been put into the common-ground (Wiltschko and Heim 2016, Wiltschko et al. 2018). Patterns of synchronic variation and change in a recent sociolinguistic corpus of Toronto English has shown that eh occurs infrequently relative to other confirmationals (e.g., Denis and Tagliamonte 2016), yet it persists as the “quintessential Canadian English stereotype” (Denis 2013: 1). This mismatch between usage and ideology leads to two questions. First, was eh more frequent in earlier Canadian English (CanE) than it is today? Eh's stereotype status might suggest a higher past frequency. Second, given Denis and Tagliamonte's (2016) focus on Toronto, how homogeneous was the variable system of confirmationals across Canada? Was eh broadly Canadian as the stereotype suggests or was it limited to particular regions? As a first step toward addressing these questions, I report on research that investigates confirmational variation in synchronic and diachronic corpora of two varieties of CanE: Southern Ontario English on the one hand and Southern Vancouver Island English on the other. While similar longitudinal data sets from more regional varieties of CanE (and perhaps other varieties) are necessary to fully answer these questions, the data discussed here allow for preliminary exploration and are intended to serve as a baseline for future research.¹

2. Background

To begin, I layout the relevant background on two fronts: 1) eh and other confirmationals in CanE and 2) dialect homogeneity in CanE.

2.1 Eh and other confirmationals in CanE

While eh has been present in CanE since at least 1836 (Denis 2013, Dollinger and Fee 2017), Avis (1972) rejects the notion that eh is a Canadianism (i.e., a word that originated in CanE or has a meaning unique to CanE) on the grounds of its presence in many other Englishes.² However, in doing so, he discusses multiple examples of eh as the subject of metadiscourse that links it with CanE. For example, he notes that Americans find eh to be “characteristic of Canadian habits of speech” (Avis 1972: 91). Indeed, Gold (2008a: 141) notes that discussion of eh as a stereotype of CanE has appeared in the literature and popular press for the last 60 years beginning with Avis in 1957 and then Allen (1959: 20) who observes its shibboleth status in contrast with American English. Gold (2008b: 74) provides many additional similar examples in the linguistics literature from the decades that follow.³

Denis (2013: 4) argues that in the mid-twentieth century, the vehicle of the enregisterment of eh was the link made to the ideological schema that place and dialect are inherently connected (Agha 2003, Johnstone and Kiesling 2008). For eh, this is played out at the level of the nation-state (Gold and Tremblay 2006: 247, Boberg 2010: 122, Denis 2013: 4, Dollinger and Fee 2017). Specifically, it is the Canadian-American cultural contrast that is critical to Canadian cultural identity; as Kymlicka (2003: 363) puts it “what defines being Canadian, perhaps above all else, is precisely not being an American” (see also Lalonde 2002, Resnick 2005).

However, the analysis in Denis (2013) is predicated on the idea that eh occurred above some threshold frequency within CanE, frequent enough to be salient to speakers and to propel through the process of enregisterment.⁴ While Avis (1972: 95) notes that “[t]here can be no doubt that eh? has remarkably high incidence in the conversation of many Canadians these days”, we have no sense of the frequency of eh relative to other confirmationals in earlier CanE. What we do know is that in contemporary urban CanE, eh is infrequent: Denis and Tagliamonte (2016) report that eh represents 3% of all “utterance-final tags” in the Toronto English Archive, which represents the speech of Torontonians born between 1916 and 1992.⁵ The variable system is dominated by you know among older speakers and by right among younger speakers; eh is low-level across apparent-time. This raises the first question: did eh occur at a higher frequency in the past, potentially aiding in the enregisterment of this feature?

2.2 Dialect homogeneity

From the earliest investigations of CanE, homogeneity (at least among urban, middle-class speakers from Ontario to British Columbia) has been an enduring feature of dialectological work (e.g., Scargill 1957). Early evidence for homogeneity focussed on lexical and phonological features (e.g., Gregg 1957, Scargill and Warkentyne 1972), but more recent sociolinguistic research has found evidence for homogeneity at the grammatical and discourse-pragmatic level in corpora of vernacular speech (e.g., Dollinger 2008, Tagliamonte and Denis 2014, Denis and D'Arcy 2019). At the same time, some of this recent work has suggested that contemporary homogeneity at the discourse-pragmatic level may have been emergent (Denis and D'Arcy 2019). This raises the second question: was eh always prevalent across Canada or was it limited to particular regions in the past?

3. Methodology

Following Denis and D'Arcy (2019), I examine two regions: Victoria (and Southern Vancouver Island) in British Columbia and the Southern Ontario region. These places represent the western and eastern edges of the General CanE dialect region (Labov et al. 2006) but both regions have similar input populations and demographics.

Settler colonial expansion of English speakers and concomitant displacement of a diversity of Indigenous peoples began in Ontario toward the end of the American Revolutionary War in the 1780s. Several English dialects (and in some cases other languages) were spoken by these early settlers but consisted primarily of American Mid-Atlantic English-speaking Loyalists from upstate New York, New Jersey and Pennsylvania (see Dollinger 2008 for a detailed language history of Ontario). Some degree of early dialect mixture must have taken place resulting in a cohesive Loyalist English which has been argued to be the ancestral variety of contemporary General CanE.

Colonialization of Southern Vancouver Island began relatively early for British Columbia; settlers arrived starting around 1843. As D'Arcy (2015: 49) suggests, although Victoria is styled as Canada's “most British city”, initial settlers came primarily from Southern Ontario and “brought with them the Loyalist legacy”. That said, the British immigrant influence was not trivial and preservation of a British gentry culture (including language) was structurally supported in the upper and upper-middle-classes (Trueman 2009: 54). While “the English model of schooling has had enduring consequences in the local linguistic ecology” (D'Arcy 2015: 50), by and large, “the result was a select social elite that did not reflect the vast majority of Victorians, linguistically or socially” (Denis and D'Arcy 2019: 227). In other words, Southern Ontario and Southern Vancouver Island share the same, ultimately American English-speaking, Loyalist settler input.

Following Denis and D'Arcy (2019), the data reported here come from corpora that allow for a two-by-two comparison: two regions and two time periods. The geographic comparison is made between Southern Ontario and Southern Vancouver Island and the temporal aspect comes by way of comparison of contemporary sociolinguistic corpora with earlier oral history data. The contemporary data come from two corpora of sociolinguistic interviews: Toronto English Archive (TEA) (Tagliamonte 2003–2006, 2006)⁶ and the Synchronic Corpus of Victoria English (SCVE) (D'Arcy 2015). The oral history data from Southern Ontario are comprised of two subsets of oral histories: the Farm Work and Farm Life Since 1890 Oral History Project ⁷ which includes interviews with elderly farmers in the Niagara Region and Eastern Ontario (Denis 2016) and the Belleville Oral History Collection (Tagliamonte 2007–2010, Denis 2015). Combined, these collections will be referred to as the Earlier Ontario English (EOE) collection. The historical data from Southern Vancouver Island come from the Diachronic Corpus of Victoria English (DCVE) (D'Arcy 2015). Details of these corpora are presented in Table 1.

Table 1: Summary of the corpora

These corpora span 113 and 131 years of apparent time for Southern Ontario and Southern Vancouver Island respectively. The 175 speakers considered for this investigation were chosen to represent a roughly uniform distribution across this apparent-time range. Speakers were also well-distributed by reported binary gender.⁸ The social class of speakers varied among the contemporary interviewees and DCVE, but by and large the data represent middle-class speakers. The social class of speakers in the EOE varies: Those from Niagara and Eastern Ontario were farmers and those from Belleville worked various, mostly middle-class, occupations.

All tokens of confirmationals were treated as ‘ways of saying the same thing’ following the variationist method. Given recent deconstruction of the multifunctionality of confirmationals by Wiltschko et al. (2018), I offer a slightly narrower definition of the variable context than Denis and Tagliamonte (2016: 90): any discourse-pragmatic feature that 1) occurs utterance-finally and 2) could be used to elicit confirmation that the addressee knows the truth of the preceding proposition. In CanE, this variable context thus includes utterance-final eh, right, you know, you see (among others) but does not include sentence tags such as isn't it and aren't you which do not target confirmation of the addressee's beliefs but rather targets confirmation of the truth of the preceding proposition itself (see Wiltschko et al. 2018: 594–595). With this variable context in mind, each speaker's interview was systematically read and all tokens were extracted.

In total, 4155 tokens were extracted.⁹ Variants included eh, right, you know (what I mean), (you) see, hey, and huh. To address the questions above, I focus on the first four variants, (see (1)), since these are the variants which each occur more than 100 times in the data.

4. Findings

Figure 1 presents eight bar charts displaying the overall proportional frequency of these four variants in Southern Ontario (above) and Southern Vancouver Island (below), and across time (the historical materials are in the leftmost column and the synchronic sociolinguistic interview corpora are on the right, split into three groups by year of birth, following Denis and Tagliamonte's 2016 groupings).¹¹

Figure 1: Proportional frequency of four confirmational variants in Southern Ontario and Southern Vancouver Island over time. Ns appear above bars.

Figure 1 begins to address both of our questions. First, eh is most frequent in the older Southern Ontario data. It represents over 12% of confirmational tokens in the EOE, four times higher than what was reported by Denis and Tagliamonte (2016). Thus, eh does indeed seem to have been more prevalent in earlier Southern Ontario English.

However, the data from Southern Vancouver Island suggests that these two regions are not consistent with respect to eh. The frequency of eh in Southern Vancouver Island, whether in the synchronic data or the diachronic data, is not higher than 1%. Thus, it seems that despite the national stereotype, there is at least one region where eh is, and has been for a long time, a low-frequency variant. That said, the other variants demonstrate remarkable parallelism between the two regions. You know and you see dominate the earlier data (with more variability in the West). From there you see obsolesces while you know briefly rises in frequency and then drops at the expense of innovative right. This pattern is consistent with the homogeneity through convergence observed by Denis and D'Arcy (2019) for general extenders, another discourse-pragmatic variable.

One outstanding issue is that Figure 2 displays the data from the EOE which collapses three regions in Southern Ontario which happen to differ with respect to rurality. This is relevant because we know that CanE homogeneity has always been presented with the caveat that parallels are found in the urban middle-class (Chambers 2004). While Victoria and Belleville are 3500 kilometres apart, they are united in their town status. Belleville was incorporated as a city in 1878 while the Niagara and Eastern Ontario areas are still today mostly rural (and certainly all of the speakers grew up in rural locales). To disentangle the possible effect of town versus country from east versus west, Figure 2 examines each Ontario location in the EOE, in contrast to the DCVE.

Figure 2: Proportional frequency of four confirmational variants in diachronic data. Ns appear above bars.

There are two important observations to make from Figure 2. First, eh is much more frequent in Belleville than Victoria. This suggests that eh was demarcated more by region (Southern Ontario vs. Southern Vancouver Island) as suggested in Figure 2, rather than by rurality/urbanity (Victoria and Belleville vs. Niagara and Eastern Ontario). Second, this seems to be true in spite of evidence for a different rural/urban split in the data. The variant you see – essentially obsolete today – distinguished town (a high frequency in Victoria and Belleville, 44% in each) from country (a much lower frequency in Niagara, 17% and Eastern Ontario 10%). I leave these data as the basis for future investigation of the question of eh, other confirmationals, rurality, and regionalisms.

5. Conclusions

I have discussed variationist evidence that suggests 1) that eh was used at a higher frequency (relative to other confirmationals) in earlier Southern Ontario English in comparison to contemporary reports and 2) that in these earlier data, the frequency of eh varied by region: higher in Southern Ontario and lower in Southern Vancouver Island. The rate of eh, at least in Ontario, was evidently sufficient for this feature to undergo the process of enregisterment described by Denis (2013). However, given eh's national status, the low frequency of eh in Southern Vancouver Island must be addressed. I take the lack of regional homogeneity not to be problematic with respect to this stereotype. Consistent with Agha's (2003) discussion of enregisterment and the valorization and spread of social meanings, eh usage and eh ideology are orthogonal. That is, metadiscourse about eh and who uses eh can diffuse through social space independently of the actual usage of eh, as evidenced by the the availability of eh commodities at tourist shops in downtown Victoria despite its near categorical absence of usage in the data from Southern Vancouver Island. As Johnstone and Kiesling (2008: 9) put it, “knowing a place means knowing its dialect”; a Victorian is not required to use or even self-identify as a user of eh in order to have the cultural knowledge that eh is Canadian. The question remains, how does this play out across the country? The data presented in this squib can serve as a baseline for investigating the earlier usage of eh in CanE during its enregisterment as a stereotype. With longitudinal data from more regions, the connection between eh usage and ideology can be further disentangled.