In Language Know-How: Canadian Perspectives in Contemporary Linguistics, ed. Rose-Marie Déchaine. 50th anniversary issue. Canadian Journal of Linguistics 50 (2005): 215-39.

Sociolinguistics and the Language Faculty

J.K. Chambers

University of Toronto

I came relatively late to sociolinguistics, after devoting my student years and my earliest professional years to abstract syntax and phonology, areas that were then called "theoretical linguistics." When I took a serious look at William Labov's initiative for incorporating the linguistic variable as a structural unit almost ten years after his first salvos (Labov 1966a, 1966b), I embraced it with a convert's zeal. One reason for its immediate appeal was that it helped me to make sense of data I had been wrestling with for a few years on the heterogeneity of Canadian Raising, about which I had published a generative phonological account a few years earlier that pretended it was homogeneous, or, actually, relegated its variability to subordinate clauses and footnotes. (On issues surrounding Canadian Raising, see Chambers 2006.)

For a converted theoretician coming to sociolinguistics, the greatest departure was that it required analysing grammatical performance in terms of extralinguistic factors as well as linguistic ones, such as communal context (the stylistic continuum from vernacular to formal) and social attributes (age, sex, ethnicity, class, and the rest). Strange as that seemed at first, there was a kind of inevitability to it if we hoped to reach an understanding of language that went beyond the combinatorial calculus of abstract, non- social theories. Sapir, the most visionary of the structuralists, had said, "Language is primarily a cultural or social product and must be understood as such. Behind the apparent lawlessness of social phenomena there is a regularity of configuration and tendency which is just as real as the regularity of physical processes in a mechanical world" (1929: 165-66). I had no doubts that I was joining Labov and others in an accelerating movement to take that next step and discover the "regularity of configuration and tendency" that would come from looking hard at language in its social and cultural contexts.

In its short history, sociolinguistics has been, by all objective accounts, one of the most productive and illuminating branches of the discipline. Its time was ripe, and with virtually everything open to discovery it made rapid strides, especially in the first decades.

Chomsky and Communicative Competence

For a linguist of my background and training, it was reassuring to see that Noam Chomsky, the most anti-social of theoreticians in his early writings, quietly revised his thinking later on, surely under the weight of evidence emerging from sociolinguistics (though he never said so). He recognized communicative competence, which he called "pragmatic competence," and gave it status as one of the "organs" of the "language faculty." For Chomsky, who had once grandly declared that "observed use of language...surely cannot constitute the actual subject matter of linguistics, if this is to be a serious discipline" (1965: 4), the recognition of communicative competence 15 years later was a significant departure. Chomsky had previously been oblivious to anything beyond grammatical competence, the system of rules and representations that underlies linguistic creativity. Now he admitted, "A fuller account of knowledge of language will consider the interactions of grammar and other systems, specifically the system of conceptual structures and pragmatic competence, and perhaps others" (1980: 92). He defined pragmatic competence as "knowledge of conditions and manner of appropriate use, in conformity with various purposes.... We might say that pragmatic competence places language in the institutional setting of its use, relating intentions and purposes to the linguistic means at hand" (1980: 224-25). In other words, it is what sociolinguists, following Hymes (1974: 75), had been calling communicative competence for several years.

Chomsky's about-face, which came at what was probably the height of his overwhelming influence on our discipline, went largely unnoticed. Chomsky has never been inclined to state explicitly that he is revising his stance on any issue. Had he done so (or had someone else drawn attention to it at the time) it might have enticed— or licensed, if you will— some of his disciples, who were often among the cream of the young scholars coming into Linguistics, to take pragmatic competence seriously, and perhaps to explore its relation to grammatical competence in the language faculty.

The inattention was not one-sided. Sociolinguists also largely failed to notice that their enterprise was now included in Chomsky's expanded model of the language faculty. So they too ignored the empirical potential for exploring interactions among the modules, starting from the communicative competence which their researches, if they really make a contribution to our understanding of language as a human attribute, must ultimately elucidate. Most sociolinguists, no less than most theoreticians, go about their business as if they are engaged in self-contained, hermetically sealed research with no implications beyond the immediate results that, say, women in Amman tend to use glottal stops talking to other women but uvulars talking to men, or that, say, ergatives and transitives subcategorize periphrastic modals but unaccusatives do not. Rightly so, in one sense. We have to know these things before we can know more.

Eventually, however, we have to know more. The next revolution in Linguistics, what might be thought of as the post-Chomskyan idea that will reform the discipline as Chomsky did in the late 1950s, might well come in the reconciliation of modules of the language faculty. As a goal of linguistic research, no one could reasonably object to it, but in the present state of our knowledge it seems well beyond our reach and hardly even foreseeable.

Coming to Grips with the Language Faculty

A preliminary step, it seems to me, is to increase awareness of the language faculty, particularly to make linguists aware that their work is intended ultimately to shed light on the language faculty. Chomsky, of course, has tried to do that in his philosophical or non- analytic writings, but he has been almost alone. It does not seem unreasonable to say that research pursuits that have no potential relevance to our understanding of the language faculty are not worth pursuing. The caveat, of course, is that we never know in advance where a research pursuit might lead us, or lead others. Making linguists more aware that understanding the language faculty is an ultimate goal of our research does not in any way define a priori which research pursuits are legitimate or worthwhile.

In this chapter, I want to discuss some research pursuits of my own that have led me to make claims about the language faculty. Those claims, I must admit, have not always been received with equanimity by my colleagues, but neither have they aroused outright hostility or, on the positive side, refutation. Each claim arose independently in the course of my work on distinctly different issues, and they come from quite disparate aspects of my sociolinguistic research spread over a couple of decades. The fact that they share this common thread„ that is, the fact that each of them (incidentally) led me to make claims about the language faculty „ went unnoticed by me (at the time) and by everyone else. In bringing them together here, as they have never been brought together before, I hope they might take on the kind of critical mass that makes a bigger impression than any one of them has made on its own.

The impetus for my claims about the language faculty did not arise from any direct attempt at coming to grips with that overarching concept. Indeed, I would not know how to go about looking directly into the language faculty. Rather the claims arose perfectly naturally (in my perspective) from the discovery that certain variable phenomena I was engaged in studying had in the end some aspect that defied straightforward linguistic or sociolinguistic explanation. At that point, I found myself hazarding an explanation that its regularities had their root in some deeper human source, the language faculty.

This line of reasoning, now that I have started looking at it self- consciously, also seems to me to be Chomskyan in its origin. One of Chomsky's abiding themes has been the distinction between, in effect, essential and accidental properties of language. Thus he talks about "conditions that are not accidentally true of the existing human languages, but that are rather rooted in the human 'language capacity,' and thus constitute what counts as linguistic experience" (1968: 24).

My research had led me, on the three occasions I will outline and discuss below, to the conclusion— or perhaps just the conviction— that some aspect of the linguistic behaviour I was studying was fundamental and immutable and somehow essential to the phenomenon regardless of linguistic conditioning or social circumstances or cultural differences or individual predilections. As such, that fundamental aspect appeared to be, as I saw it, a property of the language faculty, or at least a reasonable candidate for it. The notions are known as the Ethan Experience (2002: 121-23), Sex- Based Variability (2003a: 139-52) and Vernacular Roots (2004). (The references cited here are to my most recent and fullest treatments of each notion.) I have written about each of them several times, and they have some currency among sociolinguists. I have never previously brought them together, and indeed it was not obvious to me in the original research that led me to propose each of them that they shared the common thread that encourages me to bring them together now. At some point I realized that all three, though linguistic and actualized in social interactions, are rooted in human cognition, which is what I take to be the basis of Chomsky's notion of the language faculty.

In the discussions that follow, I am mainly interested in outlining the basis for my claims about the language faculty without recapitulating all the detail and the arguments of the original research. I start with the Ethan Experience— and a qualification. Of the three notions, the Ethan Experience is the simplest, and thus the right place to start. It is also the one that is least substantiated by empirical evidence, and in that sense the most mysterious (though all three are, in a very positive sense, mysteries). The Ethan Experience is empirically the weakest of the three perhaps through a failure of imagination (mine and others) about what might constitute an empirical test for it. But I have no doubt that it is real, and testable.

The Ethan Experience and the innate accent filter

In cosmopolitan communities today, social stratification is marked linguistically not only by class accents and dialects, which were almost the sole source of communal linguistic variation in the nineteenth century, but also by second-language accents and dialects, and often by foreign languages which persist in neighbourhoods well beyond the lifetimes of the immigrant founders. In Toronto, for example, 40 percent of the populace speaks a mother tongue that is neither English nor French, according to the 2001 census, when the population was nearly three million. The foreign-language population is not transitory, the result of a sudden swell that will assimilate linguistically by the next census. Twenty- five years ago, in the 1976 census, more than 30 percent in Toronto spoke an unofficial language natively, and in 1996 the proportion was almost 37 percent. Toronto is an extreme case now with its multilingual diversity but it is perhaps the wave of the future. In technologically developed nations everywhere, most urban areas are multilingual and multicultural, and increasingly so. Large-scale movements of people across national boundaries for purposes of permanent settlement are an irreversible consequence of global mobility. For linguists, immigration presents unique situations of dialects and languages in contact with boundless opportunities for studying linguistic accommodation and integration, to the extent that a sociolinguistics of immigration is unfolding as a branch of study (Chambers 2003b).

In societies where immigration is rife, it is the common experience of thousands of children of immigrants that they will grow up speaking the local dialect appropriate to their age and class with native fluency. Under normal circumstances, their dialects will be indistinguishable from their playmates and classmates who come from households in which the parents are not immigrants but are themselves speakers of the local dialect for two or three

or four generations or more. The linguistic difference between immigrant and native parents is thus neutralized in the offspring. As it happens, the neutralization is even more thorough than might be expected. It is the common experience of the children of immigrant parents that the foreign-accented English of their parents exerts no influence whatever on the English of their children. This observation came to my attention through a colleague at the University of Toronto, a cultural anthropologist born and raised in the Czech Republic as was his wife. Their son Ethan was born after he and his wife immigrated to Toronto, and the careful parental scrutiny of the father, a trained linguist, led him to a surprising observation. At no time, he noted, did Ethan's English show any traces of the parents' foreign-accent features, not even in isolated words.

Ethan's parents have, by their own admission, medium-to-strong accents. Among the features of their accent, they usually have, for instance, tap /r/ intervocalically, and they have relatively tense vowels especially in the KIT and DRESS series, that is, the Canadian non-low front lax vowels. Ethan showed neither of these features in his speech, not even momentarily. He never tapped the /r/ in hurryor any other word, or pronounced ship as if it were sheep and red as raid, though his parents did.

That observation is revealing in its own right, but it is reinforced by an equally revealing corollary. Ethan, though an active and academically bright child (eventually he went on to earn a graduate degree), was well into his school years before he realized that his parents spoke foreign-accented English. It dawned on him suddenly, sometime in the years leading up to puberty, though the evidence had been copiously before him from birth.

I knew that Ethan's experience was not unique to him, but I knew it only vaguely, probably from snippets reported by students from similar backgrounds over the years. The eye-witness account of a trained and sensitive observer like Ethan's father added a new level of confidence. I began lecturing about it and eventually wrote it up (in 2002) as the Ethan Experience. Naming it for Ethan quietly credits his role in bringing it to consciousness, but from the beginning I emphasized its universality. The Ethan Experience is in fact the common experience of thousands of children from multilingual homes who assimilate the linguistic norms of their home community. If I was tentative in the beginning about its universality, that tentativeness was quickly dispelled. In large classes where more than half the undergraduates in the hall fit Ethan's profile, I saw the shock of recognition sweep through the room as I described the Ethan Experience. In a course on the sociology of language, most students chose to write about the Ethan Experience for their first-term essays (under quaint titles like "The Gabriela Experience," or "Rocco's Ethan Experience") to the point where the teaching assistants suggested that we leave it off the list of topics so they could have a little variety in their reading. The Ethan Experience elicited exactly the same reaction in lectures far away from Toronto and from Canada. The published version, though less immediate, has also evoked the shock of recognition. I have come to expect it, and know to watch it happening.

Like Ethan, countless children from multilingual homes never acquire their parents' foreign-language accent features even in isolated words, and also like Ethan these children fail to register the foreign-ness of their parents' accents until late in their school years. Because it is so general and so common, the Ethan Experience constitutes principled behaviour that needs to be accounted for in theories of language contact. It does not appear to be learned behaviour, because there are no gradations of competence or slip-ups or hypercorrections. Rather, it is an attribute that is simply available to children, apparently from birth. It is well beneath consciousness, and rises into consciousness only around puberty, when the children realize that their parents talk different from them. The timing adds another factor to its cognitive significance. The dawning is presumably related to the critical period around puberty, when most of the innate equipment for language acquisition begins ossifying.

At the core of the Ethan Experience is an innate accent-filter. Immigrant offspring apparently come equipped with a device for screening out accent features that are odd and eccentric and (literally) outlandish in terms of the communal norms. It appears to be operative for grammatical features as well, that is, for features of morphology and syntax, but its importance is not as clear. One of the characteristics of Ethan's parents is that their grammars are mostly standard and their speech is fluent, notwithstanding the phonetic interference in their accents. In the beginning, I assumed that a high level of fluency was requisite for the Ethan Experience on the common-sense assumption that gross discrepancies would automatically draw attention to parental differences, but numerous students have indicated that fluency in the parents' generation may not be crucial, or perhaps even important. However that may be, the accent-filter operates at the level of phonology with astounding efficacy, so that for Ethan, for instance, pronunciations of tap /r/ were simply perceived as retroflex, tensed vowels were heard as lax, and so on.

The function of the innate filter seems clear. It effectively shields children from acquiring phonetic features that have no status in the communal language norms. If, instead, children acquired their parents' foreign accent-features, they would find themselves enmeshed in garden-path learning. The children would have to pass through unlearning and relearning sequences en route to native competence in the home dialect, and cyclic corrections of this kind have never been reported in any normal developmental studies. The acquisition path is thus greatly facilitated. Instead of acquiring 'outsider' phonology and then eliminating it, children simply fail to hear it. And by failing to hear it, they are led to believe that their parents' speech is no different from their own.

The Ethan Experience is not restricted to immigrant situations. Even if there were no evidence, this would follow as an inference from the fact that, as a property of the language faculty, it must be activated, at least potentially, in any acquisition scenario, and not just for immigrant offspring. Kirk Hazen (2002) is working with West Virginia families that have crossed dialect boundaries in their formative years, so that parents and children are speakers of regionally distinct varieties. In interviews, younger children in those families maintain that everyone sounds the same. The Ethan Experience may even filter out language differences, so that children who spoke the immigrant language to grandparents but never spoke it under other circumstances recollect their childhood experiences as taking place in the language of their parents rather than their grandparents (discussed further in Chambers 2002: 122-23).

We are now aware of situations in which the Ethan Experience can be blocked. A well- known New Zealand linguist retains a vivid childhood memory that came about when she was placed at age five in a private school usually reserved for upper-crust families. She brought two classmates home after school, and the little girls mocked her father's working-class accent mercilessly. As the linguist told me, "I was robbed of my Ethan Experience by those snotty little girls." From age five onwards she had no illusions about the phonetic disparities in her household, or about their social evaluation. Acute social embarrassment can obviously override the innate accent-filter.

A more complicated case, reported by Mariam Ahmed in an undergraduate project for a course I taught at University of York, England, involves bidialectalism. In the neighbourhood Mariam grew up in in Bradford, a large industrial town in Yorkshire, most people are ethnically Pakistani. The grandparents and parents are immigrants, but Mariam's generation are natives. In Mariam's family and apparently throughout the community, the children speak to their elders in Pakistani-accented English, but they speak to outsiders in a northern England variety that is indistinguishable from the accents of their non-ethnic friends and classmates. The differences in the two accents are sharp, and sometimes absolute. For instance, the accent they use with their elders includes retroflex consonants, which are completely absent with their peers; and the peer accent includes T-glottaling, characteristic of northern England, which is mostly absent in the domestic accent.

Solidarity is a motivating factor. As Mariam says, "There is a great deal of respect for elders in Asian culture." However, it is not clear that it is the Pakistani-accented variety that Mariam's generation is accommodating. Neither Mariam nor any of her subjects could remember becoming bidialectal, which means either the two accents developed simultaneously, or if they developed sequentially they were so spontaneous as to be indistinguishable in terms of nativeness. Simultaneous or sequential, the situation seems to pose a dilemma for the postulated innate accent-filter. Instead of inuring the acquirers to features that would be false steps in the acquisition of the standard accent, these learners appear to acquire a double standard, one for domestic use and one for the world outside.

Though the situation is complicated, it seems to me that the Ethan Experience is nonetheless involved. The two accents of Mariam and her peers develop independently, without self-conscious monitoring or interference, until the critical period, some time in the later school years, when it becomes obvious, at least to those who think analytically about their speech (as do linguistics majors), that they do in fact command two dialects for different settings and have always done so. The Ethan Experience for Mariam and her peers, instead of blocking foreign-accent features, seems to be actualized in a kind of compartmentalization of accent-features, so that both accents can develop without interference from one to the other. Though the situation is unique, the result is familiar: the acquisitional path is direct for Mariam and her peers just as it is for Ethan and his peers. The difference for Mariam is that she is acquiring two accents, but the filter segregates them so as not to require unlearning from one accent to the other. The accent- filter prevents acquirers from importing 'foreign' features into either accent.

The innate accent-filter at the core of the Ethan Experience appears to be one module of the fabled competence of children as language acquirers. We are used to hearing accounts of miraculous feats of acquisition, such as children bursting into a new language after a few hours in playschools where they are the deeply engaged in communal activities, but so far we have made few steps in understanding how those feats are accomplished. Perhaps the innate accent-filter will ultimately be subsumed under some broader property of the language faculty that accounts more generally for acquisition. So far, the best evidence for it comes from the Ethan Experience. Devising research programs to put it on firmer empirical ground may move us a step closer to the great mysteries underlying the most fundamental attributes of the language faculty.

Sex-Based Variability and sociolinguistic competence

By the time I conducted my first large-scale sociolinguistic survey with both female and male subjects in 1979, there was already enough accumulated evidence from surveys in New York, Detroit, Norwich, Montreal, Belfast, and many other places that I could make an inference— an empirical leap really— about the sex differences that emerged in my results. I was studying a variable in Canadian English called (aw)-Fronting, in which the onset of diphthong /aw/ is usually back for older people but often front for young people. My results showed that females fronted the onset vowel more frequently than males in all age groups (12, 22, 46 and over). From that, I drew the conclusion that (aw)-Fronting was a standardizing change, that is, that the fronted onset would become the norm in Canadian English.

The inference about the change being toward standardization would have been impossible ten or twelve years earlier, in fact unthinkable. But one of the great unanticipated results of sociolinguistics from the beginning is that women are always leaders in changes that replace a local variant with a more widespread variant. A few years later, that result was so solid that I could use it as a kind of lemma for interpreting my results. (The fronted onset has indeed become the Canadian norm, as shown in subsequent research summarized in Chambers 2006.)

More than that, sociolinguistic research has consistently shown that women use more standard variants than men of the same class in the same social contexts, that is, the men who are their brothers and husbands. They also command a wider stylistic range and use fewer stigmatized variants than those men.

These results, so consistent in radically different situations and distant places, cry out for an explanation. There had been a few attempts, but all of them were unconvincing. For one thing, all of them proceeded from the assumption that these differences reflected disadvantageously on women— that women hypercorrected compared to men, that they were more meticulous about their speech as they were about their dress and make- up, that they were showing up men by talking better in a way that could not be seen as an overt challenge. (These attempted explanations and other matters outlined in this section are discussed in greater detail in 2003a: 143-53.)

It seemed clear to me that differences between women and men indicated not a female defence mechanism but rather a sociolinguistic advantage. Objectively, the differences between men and women pointed to greater sociolinguistic competence on the part of women.

In the clearest cases, where the social conditions allowed for unmistakable interpretations of the ways in which language was deployed by women and men, there was no doubting the advantage of women. Meticulous studies such as Wolfram's on African Americans in inner-city Detroit (1969) and the Milroys' on working-class Belfast (Milroy and Milroy 1978) made this eminently clear. Though those studies were carried out an ocean apart with ethnically different populations, both involved the lower reaches of the social hierarchy. These are the social groups that assign the most rigid gender roles. (Actually, the highest social groups, people with inherited wealth and aristocratic privileges, also tend to make and enforce rigid roles for women and men; the true generalization about social class and gender roles is that the middle classes, the overwhelming majority in developed countries, leave female and male roles relatively vague and consequently fluid.) In both working class Belfast and inner city Detroit, men stay relatively close to home. They tend to work together, or with peers, in isolated groups in factories or shipping docks. Young men hang out in gangs in both cities, and older men tend to congregate in pubs and clubs and other recreational facilities. Women have greater geographic range, working as cleaners or clerks in office buildings and shops, traveling into central districts for shopping and movie-going, and generally getting around. In both working-class Belfast and inner-city Detroit, it is the women who represent the public face of the family, writing notes excusing the children's absence from school and speaking to the landlord or the bank manager or the children's teachers when the need arises.

These differences in female and male speech follow naturally from what I have called Gender-Based Variability. Since the women have much wider contacts, range further in their activities, and interact with a broader social spectrum, it is hardly surprising that they use standard variants as well as neighbourhood ones, or that they command a greater range of styles than do the men. So compelling is this relationship that we would expect to find the corollary in the opposite social situation, viz., we would expect men to have the sociolinguistic advantage over women in a society where they ranged further afield, mixed more often with people of higher classes and took responsibility for speaking on behalf of the family. It seems perfectly natural, of course, but then comes the disquieting revelation that in all social situations where gender roles are sharply distinguished, it is always the women whose roles encompass what might be called the linguistic chores. Gender-Based Variability, in other words, is asymmetrical, and that fact provides the first clue that it is not socializing and speaking publicly that gives women their sociolinguistic advantages, but vice-versa, that in fact it is their sociolinguistic advantages that determine their roles as socializers and speakers.

When we turn our attention to the middle class, we see that gender roles are much less well defined. In New-World societies for many generations and increasingly elsewhere, women or men can both work as accountants and lawyers and school-bus drivers, take the family car in for repairs, attend parent-teacher night at the children's school, phone the pizza delivery number, mow the lawn, or any number of other activities. The only roles that definitively distinguish women and men are those dictated by sex differences, such as child-bearing for women and physical slogging for men (digging a ditch, moving an appliance, shoveling snow). In contemporary cultures, many physical activities are automated, with the result that even the slogging admits exceptions fairly readily.

So in middle-class society, gender roles are barely differentiated at all, and there is no reason for positing Gender-Based Variability as a determinant of sociolinguistic differences. And yet exactly the same advantages show in women's speech— they use more standard variants and fewer nonstandard ones, fewer stigmatized variants, and command a wider stylistic range than men of the same social class and region.

When we seek an explanation for sociolinguistic advantages in the absence of gender- based social differences, one of the areas with some potential is sex differences. Whereas gender differences are social and get imposed on individuals by societal and cultural expectations, sex differences are biological. They show up soon after conception, with the development of differentiating genitalia, and continue developing through puberty. The most obvious sex differences are physiological, giving rise to the statistical tendencies for men to be on average taller, more heavily muscled (and therefore better suited to slogging), more likely to be left-handed and have bigger feet than women. They can be hormonal too, resulting outwardly in men's greater tendency toward whisker- growing, baldness, and kidney stones. Among the consistent sex differences are men's greater susceptibility to illnesses of nearly all kinds from the cradle to the grave, resulting in proportionately fewer stillbirths for female babies and longer life expectancy for women.

Some sex differences have linguistic implications. Men's voices are usually lower in pitch as a result of lengthening of the larynx at puberty (the "Adam's apple"). Some differences go deeper and resist easy explanations. For instance, women are less likely to stutter, or to have reading disabilities, or to lose their speech as a result of stroke or traumatic head injuries. Differences like these allow the inference that linguistic processing and production is not encoded neurologically in exactly the same way in women and men. The same inference can be drawn from the fact that in standard tests for many years women have scored higher than men on average in tasks that measure what is called verbal ability, including tests of analogy, fluency, listening comprehension of spoken and written material, vocabulary, spelling, speaking and sentence complexity. The differences have shown a statistically significant advantage for women on Graduate Record Examinations, Scholastic Aptitude Tests, Wechsler Adult Intelligence Scale, and other standardized tests taken annually by large numbers of women and men of many different nationalities.

These differences are not as well known as they might be because institutions and governing bodies have chosen to suppress them. For example, Graduate Record Examinations last published comparative statistics on females and males in 1974 (when scores on the verbal portion averaged 503 for women and 493 for men, narrowly different as raw scores but statistically significant for this large sample). Differences like these are suppressed because they are susceptible to misinterpretation unless it is clearly understood that they merely indicate innate predispositions. It would be misguided and possibly criminal if, on the basis of these facts, some authoritarian body tried to impose educational streaming or social segregation or any other divisive measures. The differences between women and men on verbal tasks amount to less than 0.5 standard deviation, meaning that the populations overlap almost completely, such that for any array of measurable skills in a woman there will be a man with exactly the same array. The differences indicate statistical tendencies and not individual profiles.

These differences are nevertheless persistent. Men's greater vulnerability to communication disorders has been recognized since the beginning of modern medicine. Women's advantages in verbal tasks have shown up in systematic testing since its inception about a century ago. Sociolinguistic research came along much later, but for over 40 years it has also shown the persistent difference between women and men in the frequency of their use of standard and stigmatized variants. Explaining the sociolinguistic differences in terms of gender roles works, as we have seen, only in social groups in which men are relatively grounded and women have wider contacts, socially or occupationally or geographically.

The sociolinguistic differences persist, however, even in the absence of sharp gender-role differences. Commonly (and increasingly), in middle-class groups in modern industrial societies gender roles overlap almost entirely. In these instances, we appear to have discovered another effect of innate female-male asymmetries, called Sex-Based Variability: the neuropsychological verbal advantage of females results in sociolinguistic discrepancies such that women use a larger repertoire of variants and command a wider range of styles than men of the same social groups even though gender roles are similar or identical.

When we find gender-based differences, as in inner-city Detroit and working-class Belfast, we are inclined to read into this a cause-and-effect relationship, with greater mobility and broader contacts on the part of women fostering development of sociolinguistic skills. It may be just the opposite. The sociolinguistic advantage may be not the result of gender roles but their cause. In societies where gender roles are sharply defined, one sex will take prime responsibility for everyday linguistic matters. It is surely no accident that it is the women who take on that role.

Vernacular Roots and the Global Recurrence of Variable Processes

The theory known as Vernacular Roots developed obliquely out of the kind of hands-on research enterprise that is the most basic work of sociolinguistic dialectologists and in some ways the most enjoyable. Some years ago I got a chance to undertake the first linguistic study of Prince Edward County, a large peninsula that juts into Lake Ontario south of Belleville, about 200 km east of Toronto. It was pioneering work in the most promising setting. Prince Edward County has a long continuous settlement history that goes back to the Loyalists in the 1790s, as do most towns on the north shore of the lake, but as a peninsula the County has been relatively isolated and rural, with fertile farmland and once-teeming fisheries. Communication routes, notably the railway in the first century and highways afterward, have reached Prince Edward County with spur lines and byways while the main routes flowed past on the mainland.

The County's isolation, which makes it especially inviting for dialectologists, has been its most conspicuous feature from the beginning. In 1851, historian William Henry Smith wrote, "The County of Prince Edward, being to a certain extent cut off from intercourse with the rest of the Province (there being but one spot at which it can be entered, except by crossing the ferries) the inhabitants, in many aspects, have scarcely kept pace with the progress of their neighbours" (1851: 257). For the locals, isolation is construed positively as the preservation of old-time values. Norm, 57, a mechanic, one of the subjects who took part in the sociolinguistic interviews, made light of their country ways in an anecdote about the corrupted innocence of the city folk. "You heard the story about the little girl who visited her grandmother, eh?" Norm asked rhetorically, and then he delivered the punch line. "The gran'mother says to the little girl, 'Did you leave that condom out on the veranda?' The little girl says, 'What's a veranda?'"

Norm was one of 18 old-time inhabitants interviewed by Craig McRae, a university student who had spent summers and weekends in the County since childhood and knew where to find the best subjects and how to talk to them. Analyzing the taped conversations, I discovered among the linguistic variables several that were well studied elsewhere in the English-speaking world. Because this was the first time that anyone had looked carefully at Prince Edward County vernacular, the discovery that many of the linguistic features were found in nearly every study of informal, casual colloquial conversations among blue-collar and farming folk everywhere in the world, led me to wonder how it came to pass that a handful of linguistic variables should turn up so persistently all around the globe.

Here are some of the common variables, illustrated with Prince Edward County examples and accompanied with lists of a few other places where the same variable has been identified and discussed. (For bibliographical references to the other studies and further discussion, see Chambers 2004.)

Other candidates for processes that recur in vernaculars all over the world are final obstruent devoicing (cubbert for cupboard) and verbal adjectives or copula absence (as in Mama sick for Mama is sick).

When I compared these variables as they occurred in Prince Edward County with their counterparts in other places, I could not escape noticing that they occurred with exactly the same variants everywhere, conditioned by the same factors (so that the final C in (CC), for instance, is everywhere deleted more frequently when a consonant follows than elsewhere), and with exactly the same social embedding (so that they occur more frequently as one goes down the social hierarchy).

An obvious question, it seems to me, is how these linguistic features came to be so widespread. Why should rural Appalachia have the same linguistic features as New York city? Or those places with Norwich or Sydney, cities separated by continents and even hemispheres? What does Carbonear, Newfoundland, have in common with Harlem or inner-city Detroit, apart from these linguistic markers in the vernaculars spoken there? What do any of these places have in common with Prince Edward County? Or Tristan da Cunha, probably the most isolated inhabited territory on the globe?

The question may be obvious, but it appeared that it had never been asked before, at least not in the living ken of my audiences and readers. The only answer with any credibility is that these features arise spontaneously, as it were, as products or by-products of some primitive mechanism of the language faculty. The alternative possibility is that they came about through diffusion, that is, the features were carried around the globe by some mediating peoples and established by contact. The diffusionist position gains some adherents because it explains the spread by appealing to external events such as migrations, social intercourse, conquests and trade, but that is also its flaw, because there is no historical record of the putative mediators. In an attempt to patch the diffusionist position, it might be suggested that these features existed in the variety that was the source for all transported varieties, but again the historical record reveals no such variety, and if it did the idea that a handful of its variable processes could have been preserved perfectly in all vernaculars for several centuries would only serve to mark them as having privileged status, thus bringing it back to the language faculty.

The clinching argument against any external, diffusion-based theory is the fact that these same variable processes occur not only in English vernaculars around the world but also in pidgins and creoles, interlanguage varieties, and— pièce de résistance— child language.

The idea that certain linguistic features have privileges of occurrence is not new. Stampe (1969: 443) proposed that standard dialects are the consequence of "an innate system of phonological processes, revised in certain ways by linguistic experience." Braine used the terms primitive and learned to distinguish processes that emerged spontaneously in child language from those that had to be superimposed by experience. In Braine's terms, "the mature [phonological] system retains all those aspects of the primitive articulatory tendencies that children have not had to overcome in the course of mastering the pronunciation of their native language" (1974: 287). Bickerton (1984) proposed the "language bioprogram hypothesis" as the cognitive basis for grammatical structures that appear to arise spontaneously in all Creole languages. Pinker and Prince (1999) compared processing mechanisms for rule-governed linguistic sub-systems with "irregularities" that require rote learning (for instance, productive verb paradigms opposed to strong or suppletive forms) with results that are interpretable in terms of cognitive naturalness of the former opposed to cognitive burden for the latter.

Obviously, languages include both the innate and the experiential (in Stampe's terms), both the primitive and the learned (in Braine's terms), both the rule- governed and the rote (in Pinker and Prince's terms). Throughout most of the history of language study, the overwhelming tendency has been to treat primitive and learned processes as if they are one and the same, undifferentiated in origin, persistence and status. For instance, everyone knows that English-speaking children pass through a period sometime in their third to fifth years in which they produce multiple negatives ("Billie didn't want none"). This homely observation is really deeply mysterious because middle-class children almost never hear multiple negatives in adult speech, certainly never from their role models, because it is never used—and is in fact highly stigmatized— in all standard varieties of English. Yet this development is accorded no more weight than other developments in child-language acquisition such as "some/any" mismatches (as in "Billie doesn't want some" instead of "Billie doesn't want any"). But the latter are idiosyncratic, unpredictable and sporadic where the former are universal, predictable and persistent. The difference is readily apparent in numerous ancillary, though no less homely, observations, among them the fact that no adult feels enjoined to 'correct' children on some/any mismatches when they use them, but adults regularly 'correct' their multiple negatives. The some/any mismatches are the result of a language- specific suppletion rule, an idiosyncratic rule in the simple sense that it has no counterpart in any other language. By contrast, multiple negatives appear to emerge in the acquisition process as a default mechanism in the language faculty. In some languages they are grammatical, but English grammar forbids them in the standard dialect. Children acquiring standard English nevertheless go through a stage where they use multiple negatives despite their utter absence in the speech of their parents, neighbours and teachers. Moreover, children go through this stage whether they are acquiring standard American, Canadian, Scots, English, New Zealand, Australian or any other standard variety, none of which have grammatical multiple negatives. Multiple negatives are primitive processes that arise spontaneously, and children must learn to suppress them as one of the inevitable steps on their way to the acquisition of the standard.

Sociolinguistics adds an important new dimension to the antinomy between the primitive and the learned because there is a clear and predictable relationship in the way primitive processes are represented in social dialects and accents. The higher the social status of the dialect, the more constrained are the primitive processes. The sociolinguistic significance was first expounded by Kroch (1978: 17-18), with specific reference to phonology:

... the public prestige dialect of the elite in a stratified community differs from the dialects of the non-elite strata (working class and other) in at least one phonologically systematic way. In particular, it characteristically resists normal processes of phonetic conditioning ... that the speech of non-elite strata regularly undergo....Dominant social groups tend to mark themselves off symbolically as distinct from the group they dominate and to interpret their symbols of distinctiveness as evidence of superior moral and intellectual qualities.

As a straightforward illustration of Kroch's point (developed at greater length in Chambers 2003a: 254-59), there is a natural tendency to simplify consonant clusters. In English, one of the manifestations of this tendency has been studied in numerous communities as the variable (CC), morpheme-final consonant cluster simplification, mentioned above. In basilectal varieties in rural northern England and Harlem African-American vernacular, and many other places, clusters can be simplified almost anywhere, whether they precede another consonant ("firs' base") or not (before an approximant as in "firs' one", before a vowel in "firs' out," or before a pause as in "I came in third, not firs'"). In varieties in which (CC) is most advanced, that is, where it occurs most freely, derived forms are sometimes pronounced as if there was no underlying cluster at all (as in "tesses" as plural of "test," and "win'y" as adjectival of "wind," that is, "windy"). Variable (CC) also occurs in standard middle-class varieties, but it is highly restricted both socially and linguistically. Socially, it occurs only in informal and casual speech, and seldom in formal circumstances. Linguistically, it occurs most commonly before a following consonant ("firs' base") and occasionally before a semi-vowel ("firs' one"), and nowhere else. In other words, the cluster is never simplified before a following vowel or pause. Phonetically, (CC) functions in vernacular varieties to simplify biconsonantal clusters, but middle- class varieties draw the line much higher and use it only to simplify tri-consonantal clusters (CC#C > CØ#C), and then only in casual speech. My contribution to Kroch's insight was to demonstrate that the sociolectal patterning applied to grammatical processes as well as phonological ones. On the basis of their social embedding, certain morphological and syntactic variables show exactly the same distributional patterns in terms of greater freedom of occurrence as we move down the social hierarchy. The purer the vernacular, that is, the more colloquial, unself-conscious, un-codified the linguistic medium, the less restricted the process. Hence the term "vernacular roots." As with any deep-seated verity, this one does not literally begin with Kroch's work. In fact, Kroch (1968: 18fn) quotes my Toronto colleague Henry Schogt making this same point in 1961: "Passant maintenant à l'examen des couches sociales d'un seul dialecte geographique, nous constatons deux forces opposées: la langue populaire riche en innovations, qui a pour elle le grand nombre, et la langue des classes aisées, qui est plus conservatrice et qui s'impose par son prestige." Modern social forces have altered the demographics of Schogt's observation, which hearkens back to the first half of the twentieth century. Urbanization, embourgeoisement, mass literacy and other factors have reduced the vernacular speakers to a minority in most of the industrialized nations.

With phonological processes, the variable typically occurs in all social varieties as we have seen with (CC), albeit less freely (or more constrainedly) in the standard. With grammatical processes, the sociolectal restrictions are sometimes absolute. Multiple negatives, for instance, are virtually absent from middle-class varieties but they occur with increasing frequency lower in the class hierarchy to the point where in New York African American adolescent vernaculars (according to Wolfram and Fasold 1974: 164- 65) negative concord in indefinite NPs is obligatory (so that "John didn't see nobody" is the only grammatical form). Such absolute differences are almost never found with phonological processes, which are typically graded so that they occur in standard speech but with tight restrictions, as in (CC). This difference follows from the well- established sociolinguistic finding that grammar divides social strata more rigidly than phonology (Chambers 2003a: 57-59).

Some grammatical processes can occur in all social varieties, and they pattern socially exactly as do (CC) and other phonological processes. Default singulars occur in standard varieties but almost exclusively when the subject of the clause they are in is expletive subject there (as in There's three students waiting for you in your office). Nonstandard dialects permit nonconcord with other subject types, in an implicational rank from they > NPpl > we > you > there (as discussed at length in Chambers 2004: 131-38). The rank order emerges with astonishing regularity in several studies that have been carried out in widely separated communities by several sociolinguists. Its order is fixed except for occasional indecision in the top two categories on the scale, such that they and NPpl (plural noun phrases) have similar frequency (so that instances of, say, They was hungry occur with almost the same frequency as The men was hungry in vernacular dialects that permit them). The order is implicational in the sense that dialects that permit nonconcord at any point on the scale will also permit it at points to their right (so that vernaculars that permit, say, We was hungry also permit You was hungry too, but not necessarily vice-versa).

Comparative sociolinguistics of default singulars in several communities around the world provides a stunning demonstration of the systematic social embedding of grammatical restrictions in stratified communities. Figure 1 graphically represents data from five communities compiled by Schreier (2002: Table 3, 85) as comparative evidence for his work on concord in Tristan da Cunha, an isolated island in the Atlantic Ocean. Schreier discovered that default singulars in Tristan da Cunha vernacular speech developed spontaneously from founder dialects that originally had variable concord marking. He says (2002: 91), "Both was and were could occur with plural subjects in the English of the immigrant population. However, the offspring of the immigrant generations, the native Tristanians, eliminated concord from the dialect and adopted invariant was regardless of subject number." Tristan da Cunha grammar, in the period when its speakers were most isolated (1859-1945), had categorical nonconcord, that is, obligatory was with all subjects, plural or singular. Evidence for their spontaneous development, needless to say, adds a cogent case for default singulars as a primitive feature with privileges of occurrence, a point that was not lost on Schreier (2002: 92): "Once homogeneity emerged and was in place, it was sustained by external factors such as immobility, isolation, and close-knit social networks.... [T]he interplay of these ... factors leads to acceleration (or in extremis, to the completion) of language-inherent changes and to the thriving of 'vernacular roots'."

In Figure 1, Tristan da Cunha is shown along the base-line. The figure represents occurrences of were with plural subjects of all kinds, that is, the occurrence of concord, in the studies made in five widely separated communities by five different linguists. Tristan da Cunha, with the pure basilectal grammar, has effectively no concord with any subject. The other four regions show increasing standard concord from one place to the next, represented graphically by the distance of the lines from the abscissa. But there is a striking serendipity in the display because the places with increasing standard concord are also places with greater social complexity, urbanization and population density. Figure 1 turns out to be a perfect graphic representation of the correlation between grammatical restrictions and social complexity. Moving up from Tristan da Cunha, the Fens are the dyked farmlands of eastern England, inhabited by agricultural workers and market-town merchants. Anniston is a small industrial city in Alabama surrounded by farmland in the southeastern United States. Sydney and York are urban centers at the core of populous cosmopolitan districts, demographically similar though thousands of miles apart. The communities in Figure 1 show wide variation in standard concord, and the variation correlates robustly, even exquisitely, with urban complexity, from Tristan da Cunha through the Fens and Anniston in the mid region, to York and Sydney at the standard extreme. The figure is analogous to a Creole continuum, with gradations from the basilect through mesolectal varieties to the acrolect, but instead of representing vernacular accents in, say, Kingston, Jamaica, these are global varieties of English.

Figure 1— Percentage of standard were with grammatical subjects in five communities (Chambers 2004: 138, using data compiled by Schreier 2002: Table 3, 85).

The detailed case study for default singulars as a vernacular universal is developed in greater detail elsewhere (Chambers 2004), as are the cases for multiple negation, (CC), conjugation regularization, and other candidates for vernacular roots. My purpose here is not to reiterate the case for any one of them (let alone all of them) but to provide sufficient detail to make the case plausible at the least and perhaps even persuasive. I began with the simple observation that sociolinguistic study of a heretofore unstudied dialect turned out to hold few surprises in terms of the range and type of variability discovered there. In fact, a number of its variable processes proved to be perfect counterparts for well-studied processes in speech communities from Australia and New Zealand to Great Britain and North America. The fact that certain processes recur in English varieties all over the world is incontrovertible. Perhaps it is because their recurrence is so commonplace that it has aroused much less curiosity than it deserves. It seems to me to be an uncanny fact, all the more mysterious because there is no plausible explanation for it in terms of contact or any other physical means of diffusion. If not diffusion, then these variable processes must arise spontaneously in their communities, and their communities girdle the globe. They might, then, provide a key to the mystery about the way in which language emerges from the language faculty which lies at the deepest reaches of the human inheritance. More practically, it seems to me that the study of these vernacular roots has implications for one of the ways in which we linguists— sociolinguists and others — might proceed in our research if it is to bring us closer to an understanding of that human inheritance.

Accidents and Essences

The language faculty is obviously implicated in the linguistic minutiae that dominate the explorations and analyses by linguists of any school. In the normal course of events, it goes unnoticed. Its invisibility might be attributed to the fact that it is either too remotely involved or too profoundly involved in what we are investigating. Even Chomsky, the promulgator of the language faculty whose conceptualization of it underlies everything I have had to say about it here (and elsewhere), has generally segregated his discussions of it to his general works. When he is engaged in actual linguistic analysis, he never invokes the language faculty. That is to say, he has never seen fit to attribute to the language faculty, at least not explicitly, such postulates as the affix shift transformation, the specified-subject condition, the root clause filter, the antecedent trace chain, the move alpha rule, the phrase-structure node little-vP, or any of the other elaborate abstractions that he has proposed and (mostly) abandoned in almost 50 years of domineering work in our discipline. It seems safe to say that Chomsky's most abiding contributions will not be his specific proposals— not the specified-subject condition and its ilk— but his ruminations on the language faculty as a species-specific, innately predetermined human endowment. It is also safe to say that the specified-subject condition and its ilk have occupied hundreds more hours by Chomskyan syntacticians than have the language faculty and other large concepts.

The language faculty, though a Chomskyan concept, may be one of the frontiers of post- Chomskyan linguistics. Sociolinguists may have an advantage by virtue of starting from more concrete foundations. Whether or not our notions of society, community, status, style, and all the other complexities we invoke as independent variables are precise and (in an absolute sense) true, they are nevertheless grounded in reality. Linguistic results correlated with, say, the sex of speakers are unlikely to be overturned or swept aside by a theory change. The accumulation of comparable correlations in several communities by different researchers over a period of years reinforces the validity of each result and increases confidence in the generality of all results. From there, it is not as audacious as it might seem to look at the cumulative results in an attempt at distinguishing what is accidental, and therefore social or cultural, from what is essential, and therefore cognitive or biological.

One salubrious consequence of coming to grips with the language faculty is that it will force the discipline outside the parochial or insular bounds that have characterized it for most of its history, sociolinguistics no less than theoretical branches. The necessity of cross-linguistic or pan-linguistic research follows from the simple observation that any attribute that purportedly belongs to the language faculty cannot be language-specific. If default singulars and consonant-cluster simplification are primitives in the language faculty, then they must also occur in other languages, presumably all languages. Putative universals based largely on English (or any other language) must have exact counterparts in other languages, other things being equal.

The discovery that women use fewer stigmatized variants and have wider stylistic range than men in the same social circumstances, the basis for Sex-Based Variability, is heavily weighted by evidence from English-speaking communities, but it is corroborated by the same findings in Japanese, Iranian and several Arabic-speaking communities (as discussed in Chambers 2003a: 153-61). Moreover, the English communities in which it has been studied are continents apart and run the gamut of social classes, age cohorts and ethnic diversity. The inference that the sociolinguistic discrepancy between women and men represents a biological difference that also shows up in female/male performances in other verbal and language-related behaviour is no more than a hypothesis, of course, but it will require a cogent alternative with solid empirical backing to dislodge it.

Unfortunately, the same confidence does not bolster the other two postulates. The innate accent-filter, as evidenced in the Ethan Experience, is reinforced by the experiences of numerous witnesses and espoused by them with a kind of fervour that initially surprised me. But the impressions of dozens or even hundreds of experiencers does not elevate their testimony above the impressionistic. We need solider empirical groundwork for it, and I suspect that putting it on solider ground might rely in the end on kindling research interest in the Ethan Experience beyond sociolinguistics, probably in developmental psycholinguistics and language-based clinical disciplines. If it happens, the rewards, as I said earlier, could provide an entry point into the cognitive mechanisms of child language acquisition, the greatest attribute of the language faculty and so far its greatest mystery.

Universalization of the theory of Vernacular Roots has hardly begun, but unlike the Ethan Experience the way of proceeding seems clear and relatively straightforward. If it were embraced by the international community of linguists, as is necessary if there is to be any headway, it could provide a bridge between variationist and categorical linguists, between sociolinguists and theoreticians, and bring a new initiative toward a unified discipline.

The way ahead basically involves language typology, a branch of linguistics that has hardly been noticed by either sociolinguists or theoreticians for some years. The basic challenge is identifying vernacular universals cross-linguistically. What do the processes share in two or more languages that allow them to be equated as the same process? What is the basilectal form, that is, the vernacular root? What constraints govern the highly restricted variety known as the standard? How are the constraints imposed hierarchically as the vernacular becomes encoded in mainstream dialects? And, ultimately, what cognitive or structure-dependent principles govern the root form and the constraints?

One prerequisite, obviously, will be fastidious studies of vernacular accents and dialects in languages other than English. Ironically, many prominent sociolinguists from Germany, Hungary, Japan and other nations have devoted themselves to the study of English vernaculars rather than their home vernaculars. As a result, there is no research bank one can appeal to on, say, Japanese multiple negatives or Hungarian default singulars or German consonant cluster simplification or any of the other putative vernacular roots.

This kind of research will impel linguists to study processes across dialects and across languages in an effort to discern what is primitive as opposed to what is learned— what belongs to the language faculty— to the bioprogram— and what is an experiential excrescence on it. Working at this depth in more than one language will require cooperative efforts among linguists with similar goals using the same terms of reference. Theoretical linguists, both syntacticians and phonologists, should play a crucial role in the comparative enterprise. They have a stake in seeking the primitives that underlie recurrent constructions in observable human speech. The research initiative seeks principles, parameters and postulates rooted not in imagination but in empiricism. In this kind of research program, linguists cannot intuit the grammatical constructions but must observe them.

Although the typological thrust that underlies these initiatives has constituted a minor and unsung branch of linguistics for many years, it was not always so. In 1963, five years after Syntactic Structures and two years before Aspects of the Theory of Syntax, in the year of Labov's ground-breaking publications on sociolinguistics, Roman Jakobson declared, "We all seem to agree that linguistics is passing from the bare study of variegated languages and language families, through systematic typological research and gradual integration, to become a thoroughly universal science of language" (1963: 275). The consensus that Jakobson announced so confidently was short-lived largely because it was curtailed by influential new developments emanating from Chomsky and Labov. Since Jakobson's declaration, the world has been transformed in all kinds of ways. Realizing Jakobson's goal is easier now in practical terms, with international mobility and global communication networks. Linguistics has also been transformed in all kinds of ways. Realizing Jakobson's goal now entails a theoretical framework that is variationist, quantitative and graded, in a sense that even a visionary like Jakobson could not anticipate. These changes have not invalidated the initiative he espoused then but in many ways have clarified it. As an ultimate goal for linguistics, his statement takes on new resonance.


Bickerton, Derek (1984) "The language bioprogram hypothesis." The Behavioral and Brain Sciences 7: 173-87.

Braine, Martin D. S. (1974) "On what might constitute learnable phonology." Language 50: 270- 99.

Chambers, J.K. (2002) "Dynamics of dialect convergence." In Investigating Change and Variation through Dialect Contact, ed. Lesley Milroy. Special issue of Journal of Sociolinguistics 6: 117- 30.

Chambers, J.K. (2003a) Sociolinguistic Theory: Linguistic Variation and Its Social Significance. Second edition. Oxford, UK, and Malden, US: Blackwell.

Chambers, J.K. (2003b) "The sociolinguistics of immigration." In Social Dialectology, ed. David Britain and Jenny Cheshire. Amsterdam/Philadelphia: John Benjamins. 97-113.

Chambers, J.K. (2004) "Dynamic typology and vernacular universals." In Dialectology Meets Typology: Dialect Grammar from a Cross-Linguistic Perspective, ed. Bernd Kortmann. Berlin and New York: Mouton de Gruyter. 127-45.

Chambers, J.K. (2006) "Canadian Raising retrospect and prospect." In Canadian English in a Global Context, ed. Peter Avery, Alexandra D'Arcy, Elaine Gold and Keren Rice. Theme issue of Canadian Journal of Linguistics X: XX-XX.

Chomsky, Noam (1965) Aspects of the Theory of Syntax. Cambridge, MA: MIT Press.

Chomsky, Noam (1968) Language and Mind. New York: Harcourt, Brace and World.

Chomsky, Noam (1980) Rules and Representations. New York: Columbia University Press.

Hazen, Kirk (2002) "The family." In The Handbook of Language Variation and Change, ed. J.K. Chambers, Peter Trudgill and Natalie Schilling-Estes. Oxford, UK, and Malden, US: Blackwell. 500-25.

Hymes, Dell (1974) Foundations in Sociolinguistics: An ethnographic approach. Philadelphia: University of Pennsylvania Press.

Jakobson, Roman (1963) "Implications of language universals for linguistics." In Universals of Language, ed. Joseph H. Greenberg. Cambridge, US: MIT Press. 263-78.

Kroch, Anthony S. (1978) "Toward a theory of social dialect variation." Language in Society 7: 17-36.

Labov, William (1966a) The Social Stratification of English in New York City. Washington, DC: Center for Applied Linguistics.

Labov, William (1966b) "The linguistic variable as a structural unit." Washington Linguistics Review 3: 4-22.

Milroy, James, and Lesley Milroy (1978) "Belfast: change and variation in an urban vernacular." In Sociolinguistic Patterns in British English, ed. Peter Trudgill. London: Edward Arnold. 19- 36.

Pinker, Steven, and Alan Prince (1999) "The nature of human concepts: evidence from an unusual source." In Language, Logic and Concepts: Essays in Memory of John Macnamara, ed. Ray Jackendoff, Paul Bloom and Karen Wynn. Cambridge, MA: MIT Press. 221-61.

Sapir, Edward (1929) "The status of Linguistics as a science." Language 5: 207-14.

Schreier, Daniel (2002) "Past be in Tristan da Cunha: the rise and fall of categoricality in language change." American Speech 77: 70-99.

Smith, William H. (1851) Canada: Past, Present and Future, Being a Historical, Geographical, Geological and Statistical Account of Canada West. Vol. II. Toronto: Thomas Maclear.

Stampe, David (1969) "The acquisition of phonetic representation." Papers from the Fifth Regional Meeting of the Chicago Linguistic Society: 443-54.

Wolfram, Walter A. (1969) A Sociolinguistic Description of Detroit Negro Speech. Washington: Center for Applied Linguistics.

Wolfram, Walt, and Ralph W. Fasold (1974) The Study of Social Dialects in American English. Englewood Cliffs, NJ: Prentice-Hall.