Counting on The Norton Anthology of American Literature

Abstract

We created a relational database that captures every author and work ever selected for The Norton Anthology of American Literature. Given this anthology's influence, our database reveals changes in the literary canon over the past half century. We find that the common story of increased diversity is true, albeit truer with respect to race than gender. However, the biggest structural change has been a substantial growth in the number of anthologized authors. We argue that, while that strategy has produced real gains, it also creates a canon that less effectively manages reader attention, affords women and people of color a less valuable position than many white male authors once enjoyed, and tacitly accepts the notion that the new additions do not have greater literary merit than authors on the original roster, whom we show editors too rarely cut.

The requirement to, in effect, create a canon (and with a Norton anthology one would always have to take this function seriously) is paralyzing enough to prevent anyone from wanting to undertake any anthology.

—Barbara Johnson, on editing The Norton Anthology of Theory and Criticism

Helen Vendler's 2011 attack on a poetry anthology edited by Rita Dove may have arrived years after the peak of the canon wars, but it recapitulated the terms of the old struggle exactly. Framed as a debate about the canon's size, it was really about the canon's inclusivity. Dove's Penguin Anthology of Twentieth-Century American Poetry, according to Vendler, makes the canon of poets too big: “No century in the evolution of poetry in English ever had 175 poets worth reading.” Vendler defends her few, preferred, and overwhelmingly white poets against the “angry outbursts” and mere “sociology” that allegedly make up Dove's new and more diverse selections. Vendler's position echoes the traditionalist response to what would have then been called Dove's multiculturalism: new selections do not belong in the canon because they have not yet “passed the test of time” (Graff and Di Leo 117). Dove responds, “I suppose Vendler would rather I declare a Top Ten, or perhaps just five, as she herself did in her recent scholarly study” (“Defending an Anthology”). Though Dove and Vendler arrived at an impasse, the scholarly consensus today seems to be that the canon wars ended with a victory for Dove's large, inclusive model, especially in US literary studies. In 2016, the editor of the inclusive Heath Anthology of American Literature remarked, “We won those wars” (Lauter, “Transforming” 31).

Dove and Vendler's exchange encapsulates a structural dispute of the canon wars: one side held that the canon of authors ought to be smaller and traditional, the other that it ought to be larger and more socially just. Jerome McGann characterizes these options as a choice between “(Right) Wisdom and (Left) Virtue,” but the implication is the same: a socially just canon would be larger and less selective than the old canon, “that White Goddess (or White Devil) of traditional (male) myth” (489, 499). A third option—a canon that is small, inclusive with respect to identity, and justified by appealing to its “literary merit” (however defined)—is scarcely ever discussed (Gottesman et al. xxiii). Few seem to have argued for expanding the canon without making it any bigger.

This framework corresponds to a profound problem with US literary canon construction in anthologies over the past half century: a strategy of authorial growth has been used instead of a more just and useful strategy of redistribution. The multiculturalists may have won the day when it comes to getting women and people of color in anthologies and on syllabi. But, as we show using a relational database that describes every selection made for every edition of The Norton Anthology of American Literature (NAAL), the result is that the canon in these anthologies has grown far larger, scattering readers’ finite attention across many more authors.¹ We argue that this strategy of growth is neither necessary nor sufficient to address the problems that editors have tried to solve with it. We advocate for anthology editors to move away from a model of canonization premised on growth and toward one premised on redistribution.

In what follows, we outline what has happened to the NAAL's canon over time through an analysis of our database. We begin by describing the relationship between anthologies and the canon wars. We then examine the consequences of the strategy of authorial growth and address the resulting devaluation of accomplishments by women and people of color. We describe how a strategy of redistribution could produce a more just and pedagogically useful canon. Finally, we emphasize that reevaluating the editorial strategies of Norton's editors is an unusually urgent matter for literary criticism, not least because these anthologies are one of the places where literary scholars wield some real power. As Gregory Jay notes, though some say “that academic politics is so nasty because so little is at stake,” when it comes to anthologies, syllabi, and literary canons, “the arguments are so fervent because the stakes are so large” (5). Of course, Norton's anthologies are a commodity produced by a business for profit, and, as a result, subject to material constraints. But they are also assigned by instructors and read by students every semester. Given their rare influence, and critics’ rare influence on them, they merit far greater scrutiny than they have lately received.

Canon Wars and Anthologies

The idea that the literary canon has expanded over the past few decades is widely accepted, but the history of this expansion remains scattered. In part, this may be because studying the canon qua canon fell out of style, though its unfashionability hardly lessened the canon's effects on literary studies. Evidence of the battles can be found everywhere, but especially in anthologies. It has long been taken for granted that literary anthologies—and specifically their tables of contents—are both an important representation of the literary canon at present and a key battleground for changing it in the future (Benedict 53–55). John Guillory writes that “we conventionally recognize as ‘the canon’ only those works included in such survey courses or anthologies as the Norton or the Oxford” (30). Leah Price asserts that “[i]n North America . . . the canon wars of the 1980s were fought over anthologies’ tables of contents” (2). And the Norton anthologist Henry Louis Gates, Jr., puts it plainly: “some of us are even attempting to redefine the canon by editing anthologies” (19). Anthologies make canons.

Through their tables of contents, the ten editions of the NAAL have made influential arguments about the form and content of the American literary canon. With its prestigious reputation, high sales, and unparalleled rate of classroom use, the NAAL surpasses nearly every other resource as a record of recent US literary canon making.² Moreover, its editions were printed over almost exactly the same period during which the canon wars were most vigorously fought: the first edition was published in 1979 and the latest in 2022. In response to criticism of The Norton Anthology of English Literature, the Norton anthologists Sandra M. Gilbert and Susan Gubar suggest that a study of “the evolving tables of contents” would allow for “something more like an adequate history” of the anthologies (1059). We have written that history for the NAAL. While Guillory is right that the literary canon “never appears as a complete and uncontested list,” our database of the NAAL's evolving tables of contents is the most detailed proxy available for the recent history of US canon formation (30).

With hundreds of authors, thousands of works, and many thousands of connections among them, our database captures large-scale trends that literary scholars have long suspected but rarely proven. For instance, the widespread notion that authors in today's anthology are less white and male than they once were is true, with important caveats that we detail below. Yet these demographic trends are all part of an overarching but undertheorized story of growth, growth that has prioritized adding new authors to the anthology instead of replacing authors already there.³ The current NAAL contains more than twice as many authors as were in the first edition. True, today's anthology covers several more decades of literary history. But that only explains some of this growth. Analyzing the evolving tables of contents shows that editors have consistently cut back existing authors’ works to make room for new authors (table 1).⁴ While rates vary from edition to edition, and the pace of growth has slowed since the seventh edition, authors are always reselected at a higher rate than works. This difference is substantial, and its effects become more pronounced over time. Put another way, NAAL editors have always more readily cut a Longfellow poem than a Longfellow.

Table 1. Number of Authors and Works Selected for and across NAAL Editions*

* On average, editors reselect 90% of authors from one edition to the next, whereas they reselect 80% of works.

This strategy of authorial growth has had two unintended consequences. First, growth undercuts one of the canon's core functions: attention management. For scholars, for readers, but most of all for students, the canon remains useful insofar as it can help them allocate their finite time and attention for reading. After all, canons emerge from the disparity between the scale of literary production and the finitude of human attention (Wilkens 249). Second, as a canon of authors grows, the diffusion of attention diminishes the impact of including new voices by giving each new author less attention than authors had received in earlier, smaller versions of the canon. And even if one argues that the strategy of growth is still viable today, it will become less and less viable over time. The 2022 edition of the NAAL contains 288 authors, but there exists some number of authors—300? 400? 100?—that is unmanageable. Though the optimal size of the canon is neither obvious nor eternal, the principle holds at any scale: As an anthology approaches the size of a library, it ceases to have value as a book.

In revealing the centrality of authorial growth to the history of the NAAL, our database also shows us something else: there are 103 authors who have been reselected for every edition.⁵ Selected for the first edition, then reselected nine times, the Norton 103 (as we have come to call them) represent what Wendell Harris calls the “glacially changing core” of Norton's canon (113). They are overwhelmingly male (76%), even more overwhelmingly white (81%), and mostly both (61% are white men). If no real surprise, this demography remains a practical problem, since instructors using the NAAL could continue to assign from this unrepresentative core while ignoring the larger, more diverse canon that now surrounds it. At the same time, the persistence of the Norton 103 in the midst of a rapidly proliferating canon points to a broader conceptual problem that the strategy of growth creates: it fits into a long history of devaluing achievements just as they are finally won by women and people of color. When Dove herself was added to the NAAL, it was structurally impossible for her to have the status that a Norton 103 member like Longfellow has. Being one of 288 can never mean the same thing as being one of 103.

The editors of the NAAL have, intentionally or not, used the strategy of authorial growth to evade the more difficult and more urgent work of redistribution. By cutting existing authors’ works rather than replacing the authors, editors have treated anthologization as a rising tide that lifts all boats. Every author newly added to the anthology is, in one sense, canonized equally. But the more times this is done, the less well it works.⁶ This is inflation not just in the sense of an expansion of the literary field but also in the sense of a devaluation of currency. Today, being an author in the NAAL is worth less than it was in 1979, whether measured in works or the more abstract but no less real unit of attention.⁷ This can be seen by flipping through the tables of contents of different editions: early editions gave whole pages of their tables of contents to writers like Henry David Thoreau, reprinting dozens of letters and journal entries along with major works like Walden. No author is afforded this kind of attention today. Through the strategy of authorial growth, anthology editors made real gains in diversifying the anthology, but, more subtly and more perniciously, inclusion in the anthology also lost value precisely because more new authors were anthologized.⁸

If editors of the NAAL were to reevaluate their strategy of growth, they would need to begin by reevaluating the Norton 103. By moving away from growth and toward redistribution, editors can make more powerful claims for the authors they add on grounds of pedagogy, social justice, and literary merit. In the terminology of the canon wars, this would mean accepting the traditionalist case for smallness as well as the multiculturalist case for social justice. In The Closing of the American Mind—the most prominent traditionalist text—Allan Bloom criticizes what he perceives as a commitment to absolute equality among multiculturalists, according to whom “indiscriminateness is a moral imperative because its opposite is discrimination” (30). Bloom's claim clarifies a common but rarely explicitly stated assumption of the traditionalist position in the canon wars—that the canon could not become more diverse without decreasing in “aesthetic value” (Lauter, “Taking” 19). We want to draw a subtle but important distinction between what Bloom got right and what he got wrong. His emphasis on the importance of what Michael W. Clune would call judgment produces a canon that manages attention well. At the same time, Bloom assumes that privileging inclusivity precludes judgment. This would only make sense if one accepted the bigoted logic that women and people of color have not produced work of high literary merit. Since this is incorrect, it both produces an inequitable canon and leaves out deserving works, whether maliciously or through ignorance.⁹ In other words, Bloom's error was not stratospheric standards but bigotry. We can judge better if we judge more justly.

Change over Time

The NAAL's texts and paratexts have affected both canons and classrooms. The NAAL is the thirty-third most assigned text in US college English classes. Aside from The Norton Anthology of English Literature, no other omnibus literature anthology cracks the top 174 texts.¹⁰ Many literature students start their university experience reading the NAAL, those who stay in the field often wind up teaching from it, and a few NAAL users have no doubt ended up on its editorial board. Its dominance makes the problems of the NAAL matter all the more. Nothing can fully represent the canon—as the former NAAL general editor Nina Baym observes, there exist “ideal canons and real anthologies”—but the NAAL's canon is a representation unparalleled in its influence.

Using our database, we can assess the aggregate impact of the canon wars on the NAAL's selections for the first time. With the help of David McClure, then technical director of the Stanford Literary Lab, we designed and populated a relational database containing every work from every table of contents from every edition of the NAAL that has been published to date.¹¹ This covers 464 unique authors and 3,374 unique works across ten editions. For every author, we record birth and death dates, gender, nationality, and race/ethnicity.¹² For every work, we note whether it is excerpted from a containing work. The result is as clear a history of the NAAL as anyone (including W. W. Norton and Company itself) has ever had.¹³

The database underpins our claim that the multiculturalists emerged from the canon wars with a meaningful, if qualified, victory, especially with respect to race and ethnicity.¹⁴ Authors tagged in our data as white in the most recent edition of the NAAL make up 57% of the whole.¹⁵ This is a substantial change from the 81% they took up in the first edition, and much closer to present-day US demographics, where the group “white alone, not Hispanic or Latino” makes up about 59% of the population.¹⁶ The proportion of Black writers has also risen from 11% to 20%, surpassing the percentage of the present-day US population identified in the census as Black (about 14%). Yet there are limits to the utility of comparisons to present-day population data, especially given the historical breadth of the NAAL. Indigenous writers are, by the standards of contemporary demography, overrepresented in the anthology, making up about 9% of authors, while indigenous people make up about 1% of the census population.¹⁷ But indigenous peoples once accounted for 100% of the population of the geographic space that is now the United States, and the NAAL is plainly not committed to including only the literature produced since the founding of the United States as an independent nation. It may be important for the NAAL to capture historical demographics that no longer obtain. At the same time, the NAAL drastically underrepresents Asian American and Pacific Islander as well as Latina/o/x authors relative to present-day demographics.¹⁸ These two groups reached their current proportions of the US population relatively recently, but that does not mean editors ought to leave writers in those groups out of the anthology to satisfy an older picture of the United States.¹⁹ Our database shows that the NAAL has been edited with an eye toward improving its demographic representativeness with respect to race and ethnicity, but this involves difficult decisions that cannot be reduced to comparisons with either historical or present-day demographics.

When it comes to gender equity in the NAAL, we do not need all that nuance. In 1979, women made up 22% of anthologized authors. In 2022, that figure grew to 36%. What is more, this number has not grown consistently: the eighth (34%) and ninth (30%) editions each had declining relative female authorship, before a rebound in the tenth. These results do not look much different if we count by works instead of by authors: in the latest NAAL, just 39% of works were written by women (and Emily Dickinson accounts for more than 18% of that).²⁰ In short, the NAAL is more representative with respect to race and ethnicity than gender, and its representation of women was in relative decline as recently as 2017. At the average pace set by the first ten editions, women authors will not reach parity with men until the nineteenth edition of the NAAL, which will feature 445 authors and be published around 2065.

These changes in representativeness on the basis of identity may not come as a surprise, though we could not have known these quantitative specifics before gathering this data. Yet the increasing racial and ethnic equity and more modest change in gender distribution are less pronounced than the trend that motivates this essay, the editorial strategy of inflationary authorial growth—that is, an increase in the number of authors at the expense of the measure of value allotted to each. The NAAL has gotten longer since 1979, increasing its page count by about 17%.²¹ But the rate of increase in the number of authors far outstrips that in the number of pages, and the number of works has actually decreased slightly. If the ratio of the number of authors to the number of pages in the first edition of the anthology had been maintained, there would be about 154 authors in the current edition. But the current NAAL has 288 authors. While the current edition contains 0.88 times as many works as appear in the first edition, it contains 2.20 times the number of authors.²² When creating a new edition, editors have consistently cut space allocated to existing authors to make room for new authors.

Once an author gets in, they have an excellent chance of reappearing in subsequent editions. On average, 89% of authors who appear in any given edition will be reselected for the following one. Slightly more than half (234, or 50.4%) of all authors ever selected have been reselected for every revision following their initial selection. In other words, once any given author gets in, they are as likely as not to never be removed. This does not quite fulfill Barbara Herrnstein Smith's prophecy that “[n]othing endures like endurance,” but it comes close (50).

While reselection is the rule, some authors do get cut. Although white men have always been the most-cut group, this is because they have always been the largest group of writers, peaking in the second edition when they held nearly two-thirds (100 of 155) of all author spots. Even though many white male writers have been cut, more have been added. The increase in white male authors over time (up from 84 to 112 authors, a gain of 28 spots) surpasses the total number of Black women authors anthologized today (25). Among authorial groups aggregated on the intersection of their race and gender, only white women (51) and Black men (33) now have more spots in total than white male authors have gained since the first edition. This holds true even though the tenth edition deaccessioned 24 white male writers, the largest such cut in the anthology's history, which partially explains how the tenth edition became the first with a lower author count (288) than its predecessor (294).

Authors’ ages have surprisingly little bearing on their selection, retention, or removal. One might expect that most churn would occur among the most recent authors as the anthology attempts to represent new literary production and scholarship. Yet, of the 333 authors added to the anthology since the first edition, only 101 of them were born after the youngest author in the first edition (Amiri Baraka, born 1934). For every recent author selected, two older ones were selected as well. If the NAAL were drifting toward the contemporary, we would expect to see the median birth year of authors anthologized in each edition tend toward the present. Instead, we see a more stochastic pattern, with a median birth year of 1877 in the first edition, the highest in the sixth (1888), and the lowest in the seventh, ninth, and tenth editions (all 1876). If the NAAL's canon of authors today seems disproportionately weighted toward the present, it has not become increasingly so. In four of nine revisions, editors selected more new authors at or below the median birth year than above it.

The relationship between literary form and anthologization is also complex. One might expect patterns within forms to mimic the overall strategy of growth—for instance, editors might cut three poems from a previously selected poet and give them to a new poet. To test this, we tagged every work for genre, using genres mentioned in the NAAL editors’ headnotes wherever possible, then aggregated the genres into a few broad forms: prose, drama, poetry, and other. Of course, the boundaries between these designations change over time, and many works defy simple classification. But taken in the aggregate, these formal designations can capture some rough trends. Perhaps most notably, over the twenty-first century, poetry has been squeezed into an ever smaller corner of the canon. The total number of poets (defined as authors with at least one poem in a given edition) grew rapidly through the seventh edition, when it peaked at 114, but it has fallen ever since, landing at 88 in the tenth edition. The total number of poems has fallen, too, from 829 in the third edition to 525 in the tenth. This change in absolute terms accompanies a decline in relative terms, since the total number of anthologized authors has grown over the same span. Poets held 47% of author spots in the first edition and hold 31% today. But while poetry as a form has lost ground, the growing ratio of authors to works observed in the anthology as a whole is evident here as well. In the first edition, poets had a median of 10 poems. This has dropped to just 4.5 in the tenth edition. Still, the change with respect to poetry does not map directly onto the overall pattern of growth, since different forms have changed in different ways. The upshot is that generic change does not follow directly from the growth strategy. Although the reduction of works per author occurs across genres, the relation between growth and the anthologization of genre is complex and merits further research.

A subtle but important implication of all this data is that the anthologized canon is defined more by authors than by works. Like many findings in digital humanities, this may feel obvious, but it has not been explicitly articulated using previous scholarly methods. Indeed, many influential analyses of literary canons focus on works as the relevant unit of the canon. Harold Bloom asserts that “texts struggl[e] with one another for survival” before “works join the Canon” (20), and Pierre Bourdieu summarizes canon expansion as the process whereby the legitimacy “acquired by frequent contact with . . . literary and philosophical works recognized by the academic canon, comes to be extended to other, less legitimate works” (26). Of course, you cannot anthologize “Song of Myself” without anthologizing Walt Whitman, so any talk of the canon will necessarily address both works and authors. Still, the evidence suggests that Norton's literary canon has moved away from what Matthew Arnold called “the best that has been thought and said” and toward a list of those people who thought and said best (viii). There is no better evidence for this than the 73 times that editors cut all of an author's previously anthologized works, then reselected that author with entirely new works.

The structural consequence of the growth strategy for authors is obvious if we analogize the anthology to a roundtable: more speakers, shorter presentations. Casting this in terms of canon inflation shows that, ultimately, the growth strategy used by the NAAL represents an editorial half measure. Even if cutting a few more obscure works to make room for a new author may appear to be desirable in individual cases, this strategy structurally devalues all the authors that those cuts were made to anthologize. Whatever boon to representativeness that strategy of authorial growth provides is a pyrrhic victory because it devalues anthologization itself.

Editing an Anthology

Anthology editors—and Norton's in particular—are hardly in the business of making aesthetic judgments free from material constraints. David Damrosch, the founding editor of the Longman anthology, notes that a new edition of an anthology “can rarely risk changing more than ten or fifteen percent of the selections previously available” or they will lose sales (“Best” 1066). More than perhaps any other work produced by literary critics, anthologies are subject to such financial pressures.

But those pressures have led to alarming quietism about the possibility of change. Such quietism betrays a devaluation of critique by a profession broadly invested in it. Sean Shesgreen has argued that the content of the literary canon “is not determined by ideologies that are debated at the MLA. It is determined by economic and material principles” (qtd. in Ayoub). Instructor surveys are the key “economic and material” factor. Norton asks instructors to identify selections from the current edition that they want for their courses: the NAAL's current general editor, Robert S. Levine, notes that “Norton does a market survey to see which authors instructors are teaching in their surveys, and which authors instructors want cut or added” (136), thereby reflecting a “‘consensus,’ derived from polling” (Johnson 113).²³ Permissions costs, which can be prohibitively expensive even for publishers like Norton, also influence editors’ decisions (L. Bloom 416; Andrews 110). At times it can seem as though these market forces completely overwhelm the critical role of editing, leaving editors to mourn their lost “fantasies of influence” (Puchner 80).

Yet despite such conservative pressures, over time Norton's anthologies are anything but stable. Only 9% of the texts selected in the history of the NAAL (314 of 3,374) have been reselected for every edition.²⁴ This suggests that Norton editors have historically had a fair amount of latitude in deciding what to include. And, measured by the number of people affected, these decisions may well be the most consequential of their scholarly practice.

Growth and Devaluation

If the history of the NAAL is marked by a preference for growth in lieu of redistribution, what distinguishes these two strategies? Redistribution tackles inequality directly by taking from the haves and giving to the have-nots. In effect it posits a “zero-sum game” in which equality can be achieved only at a cost to the most powerful (Andrews 109). Growth sidesteps this political problem by attempting to give the have-nots more without requiring the haves to lose as much. This is one influential conception of what has happened in the United States economy during the period in which the NAAL has been published: the census shows that incomes grew even for poor households, but inequality of wealth skyrocketed (Piketty 24). For most people, the modest increase in their income might have been more visible than the massive gains of the already rich, meaning growth obscured (and, arguably, enabled) an increasingly unjust apportioning of resources. In the NAAL, the case is not so extreme; some authors have been cut, and long-standing authors have lost works. But the preference for growth instead of engaging the politics of redistribution holds.

As it only partially rectified inequalities, so too has the growth strategy devalued what it means to be an author in the NAAL's canon, whether measured in works or attention. And because the changes to the NAAL author list clearly reflect an attempt to rectify historical racism and sexism, this structural devaluation is all the more troubling. Since 83% of the women and people of color who have ever been in the anthology were added after the first edition through the strategy of authorial growth, the NAAL editors have, however unwittingly, participated in an ugly tradition.

As we noted earlier, traditionalists in the canon wars often deflected accusations of bigotry by appealing instead to judgments of literary merit. Harold Bloom, for instance, laments that “the ‘expansion of the Canon’ has meant the destruction of the Canon, since what is being taught includes by no means the best writers who happen to be women, African, Hispanic, or Asian, but rather the writers who offer little but the resentment they have developed as part of their sense of identity” (7). This argument has influenced some of Norton's editors as well. M. H. Abrams, a former editor of The Norton Anthology of English Literature, complained that he had “not found ten lines worth reading in any of the women added. . . . People want these [women] but don't use them. And we have to put them in to be p.c.” (qtd. in Shesgreen 296–97). Bloom and Abrams both object on purported grounds of literary merit, proclaiming what they consider to be “worth reading.”

Such arguments may be better understood as examples of what Koritha Mitchell calls “know-your-place aggression”: “Any progress by those who are not straight, white, and male is answered by a backlash of violence—both literal and symbolic, both physical and discursive—that essentially says, know your place!”²⁵ When people who are not straight, white, and male achieve something, “violence pursues them because they accumulate achievements, and American culture is designed to remind everyone that accomplishment is meant for straight white men” (253–54).

Strategies for excluding people who are not white men often fall along two lines. In the case of African American history, “When Black people rose, racists either violently knocked them down or ignored them as extraordinary” (Kendi 125). In the first case, we see the know-your-place aggression that Mitchell identifies. By excluding indigenous writers entirely, the first edition of the NAAL took something like this approach, effectively barring writers such as Zitkála-Šá at the gates. By anthologizing just two Black women in the first edition, the NAAL exemplified something closer to the second strategy, a form of tokenism in which the exception can be construed as proving a rule.

Though exclusion and tokenism are two of the most common tactics for reminding everyone that “accomplishment is meant for straight white men,” there is a third possibility: When achievement in a field is undeniable, devalue the field. As more women enter an occupation in the United States, for example, wages for that occupation decrease (Levanon et al. 876). This logic extends to cultural capital as well. Sara Ahmed relates the following example: “When a colleague of mine, a feminist of color, became a professor, someone said to her, ‘They give professorships to anyone these days.’” As Ahmed puts it, the logic of field devaluation is that “the very fact of your arrival erodes the value of what it is that you enter . . . progression becomes deflation” (147).

The NAAL's growth strategy has acknowledged the undeniable literary merit of women and people of color while devaluing literary canonicity itself. By using growth to avoid the conflict inherent in redistribution, the NAAL editors have inadvertently replicated the devaluation strategy. If the initial selections in the first edition of the NAAL were biased, it makes little sense to correct them by, effectively, leaving them in place while affording women and people of color increasingly devalued spots on the periphery of an inflated canon. Justice and the recognition of literary merit in the zero-sum game of producing a canon small enough to manage attention well demand excising more of the white male authors selected in that biased process, so that the space afforded to the women and people of color who have always deserved to be in the canon will have the value it ought to have.

Why Not Disavow the Canon?

Before discussing the strategy we proposed in the introduction—“expanding” the canon without making it any bigger—it is worth considering two other approaches to addressing persistent inequities in the NAAL canon. First, there is the possibility of disavowing the canon altogether. Scholars have long made the case for this approach, not least because “the drive to reformulate the canon bespeaks a conservative hankering to restabilize” what feminist and antiracist critiques successfully destabilized (Buell 103).²⁶ To the extent that the NAAL's canon reflects the United States, it will inevitably reflect the “barbarism” that Walter Benjamin saw in every “document of civilization” (256). “Epistemic violence” is one such barbarism (Spivak, “Can the Subaltern” 35–37). A canon necessarily privileges some things at the expense of others. Given that the inequalities of the past influence those of the present, it is fair to wonder whether canons inevitably reproduce harm.²⁷

Without denying the validity of this critique, we cannot avoid three conceptual difficulties that disavowing canonicity entails. First, literature is longer than life. Thousands of new novels are published in English annually, but most Americans report reading five or fewer books per year (Faverio and Perrin). The question is not whether readers’ attention will concentrate on some subset of literary production (Fowler 97). It is only how we, as literary scholars, assume responsibility for guiding that finite attention. Second, if we disavow the canon just as it begins to better represent women and people of color, we risk reinscribing the erasure that the canon wars challenged in a process akin to one Gates identifies: “precisely when we (and other Third World peoples) obtain the complex wherewithal to define our black subjectivity in the republic of Western letters, our theoretical colleagues declare that there ain't no such thing as a subject, so why should we be bothered with that?” (36). Theorizing away hierarchy at the moment women and people of color gain power is just another way to confront undeniable merit by devaluing a field, and disavowing the canon now evades redistribution yet again. Third, anthologies like the NAAL exist and will be revised as long as they continue to be profitable. Given their unique status in universities, Norton's anthologies will likely continue to play an outsized role in directing attention for both instructors designing syllabi and new generations of readers. David Palumbo-Liu observes that, despite the risk of the academy's “co-opting the cultural objects of heretofore marginalized peoples” through their incorporation into literary canons, scholars recognize “the necessity to carry on even in the face of such possibilities” (3). Whether or not the literary canon ought to exist, the NAAL does.

One might object here that other anthologies create canons that supplement the shortcomings of the NAAL. Indeed, Norton itself offers anthologies that construct many different literary canons, ranging from the maximalist (e.g., world literature, poetry) to the highly specific (e.g., English Renaissance drama, American women regionalists). Focused anthologies like these have done important work strengthening literary traditions (Kinnamon 468–69; Chon-Smith 41). Anthologists such as June Jordan, Audre Lorde, and Toni Cade Bambara “acted on a conviction that authorship . . . is widely distributed despite cultural institutions that privilege the voices of a narrow, white, male elite” (Savonick 37). To give a specific example: Tara Fickle calls the pioneering 1974 Aiiieeeee! An Anthology of Asian-American Writers “Asian American studies’ controversial origin story,” noting that, in spite of its serious shortcomings, the anthology represented a major achievement in the context of racist resistance from critics and a dearth of attention to Asian American literature (xx). Similarly, upon the release of The Norton Anthology of Latino Literature, Kirsten Silva Gruesz finds herself “as curious as anyone to see what effects the Norton branding may have on popular conceptions of Latino culture” (336) suggesting, as Gates did of The Norton Anthology of African-American Literature, that in producing such anthologies under the sign of The Norton Anthology “we are canon makers and canon breakers at the same time” (qtd. in Gruesz 338). Anthologies can forge and strengthen literary traditions.

But such anthologies cannot solve the problems facing the NAAL because they do not claim national representativeness.²⁸ The fact that women and people of color remain underrepresented in the preeminent national anthology perpetuates the situation that Toni Morrison described in 1989: “There is something called American literature that, according to conventional wisdom, is certainly not Chicano literature, or Afro-American literature, or Asian-American, or Native American, or . . . It is somehow separate from them and they from it” (368; ellipsis in source). The issue is not whether students can purchase a second anthology in order to get enough writing by women or people of color. It is whether they can find their America in the one huge book they had to buy for class.

Reshaping the Canon

The strategy of growth has made durable gains in representation. It is no small thing for teachers and students to have easy access to writers like Frances Ellen Watkins Harper, Sui Sin Far (Edith Maud Eaton), and Leslie Marmon Silko, and these achievements are especially evident in the tenth edition. That the strategy of growth has been continuous across nearly every edition represents a major, sustained victory for the multiculturalists. But having claimed this victory, it is time for another phase in which we acknowledge the power of the structure that the traditionalists advocated while rejecting the bigotry that pervaded it. An inclusive canon that better adjudicates literary merit is possible.

The Norton 103 give us the clearest possible grounds for change. This group represents an unbroken continuity with 1979, with its 63 white men overwhelming its women (25) and people of color (13 African Americans, zero indigenous people, zero Asian Americans or Pacific Islanders, and William Carlos Williams the only Latino person). Cutting authors from the Norton 103 would serve a redistributive function by explicitly reckoning with the choices from 1979 that contemporary editors and instructors have chosen to reaffirm while reevaluating a demographically unrepresentative portion of the canon. This would be transformational, changing the model of authorial canonization from one of growth (new authors join a canon growing in size) to one of redistribution (authors survive in a stable or shrinking canon).

This could be easier than it might seem. First, there are many authors in the Norton 103 who have drifted into what Damrosch has called the “shadow canon” (“World Literature” 45). John Winthrop, John Greenleaf Whittier, Joel Chandler Harris, Randall Jarrell, A. R. Ammons, and Hamlin Garland are all there. However important their work might be, they hardly seem indispensable in the American literary canon today—if the preceding sentence had said “are not there,” would it seem more or less surprising to you? Scholars do write about these authors sometimes: the MLA International Bibliography shows that Whittier has been the subject of 33 scholarly articles since 2000 and Ammons of 92, with the rest scattered somewhere in between. But this is commensurate with someone like Ida B. Wells (56) or María Amparo Ruiz de Burton (85), and far short of authors like Louise Erdrich (360) or Zora Neale Hurston (621). Moreover, it does not seem as if many instructors would object to their removal—only one text by one of the white male authors listed above cracks the top 1,200 most frequently assigned texts in college English classes, and it is in 1,172nd place.²⁹ In a world of limited attention it makes sense to redistribute space away from Harris and to someone like Hurston (Their Eyes Were Watching God is the fourth-most assigned American novel in colleges). Cutting authors could make the NAAL more representative of the demographics of the United States and of the material that scholars research and teach.

If removing authors feels like a painful break with tradition, it is helpful to remember that the tradition is neither as old nor as stable as it might seem. The roster of authors that many contemporary scholars of American literature identify as canonical did not take shape until the early twentieth century. Herman Melville is the best-known example—Moby-Dick was out of print for years, and as late as 1925 a scholar found that the New York Public Library's copy of his long poem Clarel had never been opened (Delbanco 287). Americanist scholars did not arrive at their “classic eight”—identified in Norton's 1963 anthology Eight American Writers as Poe, Emerson, Thoreau, Hawthorne, Melville, Whitman, Twain, and James—until mid-century, and even someone as canonical as Dickinson did not fully enter their ranks until as late as, arguably, the 1980s (Foerster and Falk; Bibb 386–430; Csicsila 17).³⁰ Even at the top, the US literary canon has taken shape within tenured memory.

The instability of the American canon is even more pronounced when it comes to anthologies by publishers other than Norton. Collegiate literature anthologies preceding the NAAL range from The Oxford Anthology of American Literature (1938) to Wiley's America in Literature (1978), and none of them looks quite like the others (Debo 443). In fact, late-nineteenth- and early-twentieth-century US anthologies often had far larger numbers of authors than their mid-twentieth-century successors (Pressman 57–58; Csicsila 15). Their disparities owe something to time, something to criticism, and something to competition among publishers. Editors of a new anthology cannot simply copy their competitors’ product, which is one reason why they so carefully guard the tables of contents of their new editions (Damrosch, “Best” 1062; Kuipers, “Anthology” 128). Tables of contents are not merely excerpts from Guillory's “imaginary” lists nor Price's “battlegrounds”; they are also selling points. These considerations collided when the editors of the Heath Anthology of American Literature challenged Norton's market dominance in the early 1990s. The Heath attacked the NAAL from the left as “part of a broader movement for racial and gender equality” (Lauter, “Transforming” 31). The Heath emerged from the Reconstructing American Literature project, the slogan for which reads “so that the work of Frederick Douglass, Mary Wilkins Freeman, Agnes Smedley, Zora Neale Hurston and others is read with the work of Nathaniel Hawthorne, Henry James, William Faulkner, Ernest Hemingway and others” (Lauter, Canons 162). The Heath project of “reading with” affirms the centrality of the strategy of growth to anthologization beyond the NAAL, and the influence of the Heath shows that this was a useful strategy for making anthologies more just than they were in 1979. Yet, as we have argued, at some point anthology editors will be obliged by virtue of sheer quantity to make the case for reading instead of. Given the NAAL's current rates of authorial reselection, reading with is ultimately a doomed project.

There are good reasons to believe that a Next Norton 103 would be radical in the etymological sense of living at the roots of America's young, always-contested canon. But it would be radical in the usual sense of the word, too, the one Saidiya Hartman uses when she says that ending the “possessive investment in whiteness . . . requires a radical divestment in the project of whiteness and a redistribution of wealth and resources.” Creating an aesthetically exclusive and demographically inclusive Norton Anthology of American Literature would require its editors to abandon growth and pursue this kind of radical redistribution. Eventually, authors who today seem unassailably canonical would be challenged. At that point, we would have arrived at the question that the NAAL's strategy of growth is designed not to answer.