1. Introduction: Canonizing Literary Texts

The practice of canonizing literary texts dates back to antiquity. It is due to processes of canonization, for example, that seven of Aeschylus’, seven of Sophocles’, and ten of the nineteen extant tragedies of Euripides have come down to us in the form of library copies, while several hundred other dramas (by these three tragedians as well as by others) are only known by their titles and by short fragments (cf. Lesky 73-74; and Gruber). The selection of these canonical works was obviously effected by one institution: school; and schools and universities have played a decisive role in establishing canons ever since.¹ From the sixteenth century onward the emerging book market joined in the processes of canon formation, as editors and booksellers made choices as to what purchasers of books would, or should, read.²

Why is it that schools, universities, and common readers prefer certain literary works over others? Which are the criteria which lead to a text’s inclusion in, or exclusion from, the canon? According to a common assumption, these choices depend on ideologies and the powers exerted by influential elites. Indeed, a few years ago I read and published a paper on “Canon Formation in British Literature Studies,” in which I argued that because of various ideological biases a small segment of nineteenth- and early twentieth-century English novels came to represent this period on university reading lists (Kullmann, “Canon Formation”). The opposite view is taken by Harold Bloom, who in his monumental work on The Western Canon claims that canons are, or should be, formed on the basis of aesthetic criteria alone. “Aesthetic value” (22), he insists, does not depend on the ideologies of the authors [→ 65] of literary texts (see 28), and should not depend on the social group the readers happen to belong to: “I myself insist that the individual self is the only method and the whole standard for apprehending aesthetic value” (22). The benefit readers derive from reading canonical works is “the proper use of one’s own solitude, that solitude whose final form is one’s confrontation with one’s own mortality” (28).

Bloom feels the need to defend the “Western Canon” against critics of the “School of Resentment” (20) who consider canons to be a discursive tool used by the ruling classes to manipulate dependents. By, however, insisting on the solitude of the reader, “the mind’s dialogue with itself” (28), Bloom deprives himself of one of the central arguments which legitimize canon formation: if we agree on reading a common set of books, we can share our reading experiences with others. The pleasure and profit we derive from reading is supplemented by the pleasure and profit we derive from discussing canonical texts with our peers, as well as with people unknown to us (see Kullmann, Reading Nevernight, esp. v-ix). This particularly applies to the plays of Shakespeare, whose literary eminence Bloom considers unmatched (see 23-24). Shakespeare’s plays were not meant to be read in solitude but to be experienced in a theatre,³ often in the company of friends but certainly as a member of a large community of spectators.⁴ Nowadays, in the twenty-first century, the shared experience of reading Shakespeare offers ways of communicating with people of highly divergent cultural backgrounds, as seen, for example, in the conferences and proceedings volumes of the Asian Shakespeare Association.⁵

To find out about possible reasons for the canonicity of literary texts I propose to resort to Shakespeare’s sonnets as a test case. While the whole sequence of 154 sonnets, first published in the quarto edition of 1609, has become canonical, certain sonnets are time and again chosen to represent the poet in anthologies, while others can only be found in complete editions. The original sequence does not accord a privileged position to any of the sonnets.⁶ Almost all of them share the same form, with fourteen lines written as iambic pentameters following a certain, [→ 66] and fixed, rhyme scheme. This being so, how come it is usually the same sonnets that are quoted, discussed, and reprinted?

In order to determine if canonization depends on timeless aesthetic qualities or on the cultural concerns of certain periods and social environments (or on both), I suggest pursuing a historical approach. Looking at the anthologies compiled in different epochs and addressed to different communities of readers, I intend to compare the selections and to suggest possible reasons for the respective choices and specificities.

This examination will then lead me to a hypothesis concerning textual reasons for canonical preferences: anthologists might prefer sonnets in which a speaker expresses his feelings over those in which he addresses another person. To substantiate this supposition, I propose to classify Shakespeare’s sonnets as to communicative categories, on the basis of Roman Jakobson’s model of verbal communication. With the help of this classification, I will demonstrate that there is indeed a regularity in anthologists’ preferences for certain forms of communication, which may be due to cultural factors, such as a community’s particular interests and predispositions, and certainly does not depend on aesthetic value alone. As a side effect, this analysis will offer a fresh perspective on some literary features of Shakespeare’s sonnets which often go unnoticed.

2. Survey of Poetry Anthologies Featuring Shakespeare’s Sonnets

A major stepping-stone in the history of anthologies is Francis Palgrave’s popular five-volume Golden Treasury, published in 1861. While there had been poetry collections before Palgrave, the modern habit of collecting English poems from various periods in anthologies proliferated after that date. For the purposes of this paper, I have consulted five early anthologies (1783 to 1860) and 33 collections published from 1861 onwards, each of which contained some but not all of Shakespeare’s sonnets, with the figures ranging from four sonnets (in Allingham’s 1860 Nightingale Valley and Whiteford’s 1903 Anthology of English Poetry) to 60 sonnets (in John Wain’s 1990 Oxford Anthology of English [→ 67] Poetry). My aim was to include most of the British and American anthologies published for studying purposes as well as for the general reader. Concerning pre-Palgrave anthologies, my list basically consists of the collections discussed in Kingsley-Smith, The Afterlife of Shakespeare’s Sonnets; concerning the post-Palgrave anthologies I proceeded from the anthologies currently found on the international market and accessible through the German inter-library loan system. I then searched prefaces and introductions for references to previous anthologies, which, in turn, informed me about earlier collections. I also included anthologies compiled in continental Europe, in Germany, Poland, and Hungary, for study purposes. This latter group of texts, found on the German book market as well as by means of internet searches, rather has the character of a random sample. The conformity of the results, however, may justify this procedure. Finally, my corpus includes recent collections which (like Palgrave’s Golden Treasury itself) are intended for the general reader: Bulbeck’s The Illustrated Book of Shakespeare’s Verse, published in 2014 by Flame Tree Publishers (mainly sold in the gift shops of Shakespeare’s Birthplace and other sights which are of interest to sightseers), and Allie Esiri’s volume titled Shakespeare for Every Day of the Year, published in 2019. Another unorthodox collection is that established by the website of the “Poetry Foundation.” As this website is open to additions, no date can be given; for the purposes of this research project I proceed from what was there in April 2023.

Altogether, 130 out of the 154 sonnets were chosen by at least one anthologist. 82 sonnets were chosen at least twice. 45 sonnets were printed five times or more, 27 ten times or more (see Appendix I). Most of the collections have an individual note, in that they include sonnets little anthologized elsewhere.⁷ At the same time, all the post-Palgrave anthologies agree on the canonicity of a small number of core texts. There is hardly a collection which does not feature sonnet 18, “Shall I compare thee to a summer’s day?”; 73, “That time of year thou mayst in me behold”; and 116, “Let me not to the marriage of true minds.”⁸

 [→ 68] In other ways as well the lasting influence of Palgrave’s collection is evident. Later nineteenth- and twentieth-century collections usually follow the Victorian anthologist in rejecting the procreation sonnets (1-17) altogether and being extremely choosy with regard to the dark lady sonnets.⁹ The stability of the canon can be assessed by the fact that seventeen of the twenty sonnets selected by Palgrave belong to the group of 27 sonnets printed in ten or more collections. There is just a small group of sonnets which were apparently popular in Victorian and Edwardian times but then fell out of favour with anthologists. This group includes sonnets 54, 57, and 109. The second half of the twentieth century saw few additions to the “Palgrave canon,” most notably sonnets 20 and 130, which had not attracted much interest before. Sonnet 20 was certainly chosen for its take on homoeroticism¹⁰ while 130 obviously tied in with the iconoclastic discourse which had been popular since the modernist movement of the first decades of the twentieth century.¹¹ The publication of Anthony Burgess’s novel Nothing Like the Sun in 1964 may well have triggered the adoption of this sonnet into the canon.¹²

To a certain extent, the Norton Anthology marks a new departure in that it strives to compress English Literature in its entirety into the book covers of two volumes, which are obviously considered sufficient reading for an undergraduate course on English Literature. The Longman and Broadview anthologies have followed suit. All of these publications give extensive coverage to Shakespeare’s sonnets, 32 of which are found in Longman, 42 in Norton, and 45 in Broadview. When we compare the selections, we note that Norton and Broadview have 35 sonnets in common, while Longman shares 27 of its 32 sonnets with Norton. 24 sonnets are found in all three anthologies; we might call them the “Norton canon,” which is only partly identical with the “Palgrave canon.” Of the 24 sonnets of the Norton canon, only ten also occur in Palgrave, so we may notice that a certain shift has taken place. New sonnets include 1, 12, and 15, as well as 35, 80, 93, 128, and 144, while 64, 104, and 146 are no longer considered essential. It appears that there is a new interest in sonnets conveying biographical information; this [→ 69] may account for the inclusion of two of the procreation sonnets as well as sonnets referring to personal quarrels and entanglements.

Our examination of anthologies can be supplemented by some statistical evidence concerning scholarly interest, which can be obtained by looking at the MLA bibliographical database.¹³ Looking for publications on individual sonnets we find that 23 sonnets are discussed in six or more publications each. This group of sonnets roughly corresponds to the canon established by the anthologies consulted. With regard to three sonnets, 73, 129, and 116, the database yields more than 30 entries each; sonnet 20, with 23 entries, comes next. What we see here is that scholarship, to some extent at least, follows, rather than sets, the canon: many of the articles on 73 and 116 were published to provide pedagogical aids to teachers and lecturers. It is obvious that these two sonnets, with their unexceptionable messages about ageing and true love, and avoidance of the issues of sexuality and the young man’s beauty, are teachers’ favourites.¹⁴ Sonnets 129 and 20, by contrast, are apparently accorded scholarly treatments for the opposite reason: “Sex sells,” and the images chosen by Shakespeare to represent the vagaries of sexual desire are often considered daring and provocative, and they have invited speculation as to the sexual practices referred to. The only significant departure of what can be called the “MLA canon” from that established by the collections lies in the fact that the two mythological sonnets concluding the sequence (153 and 154) are accorded extensive scholarly treatment while they are generally neglected by anthologists.

3. Types of Address and Reference

In order to determine possible reasons for these preferences, I propose to start from that sonnet which on the evidence of extant manuscript copies was a seventeenth-century favourite: sonnet 2.¹⁵ Sonnet 2 was one of the eighteen sonnets included in George Kearsley’s volume titled The Beauties of Shakespeare, Selected from his Plays and Poems (1783). It was also the only Shakespearean poem included in the three-volume English Anthology, published in 1793, the focus of which was contemporary, i.e. [→ 70] eighteenth-century poetry. Out of the modern (i.e. post-Palgrave) corpus of 33 anthologies only three (Chambers, Jones, and Broadview) include sonnet 2:

When forty winters shall besiege thy brow,

And dig deep trenches in thy beauty’s field,

Thy youth’s proud livery so gazed on now

Will be a tattered weed of small worth held:

Then, being asked where all thy beauty lies,

Where all the treasure of thy lusty days,

To say within thine own deep-sunken eyes

Were an all-eating shame, and thriftless praise.

How much more praise deserved thy beauty’s use

If thou couldst answer ‘This fair child of mine

Shall sum my count, and make my old excuse’,

Proving his beauty by succession thine.

This were to be new made when thou art old,

And see thy blood warm when thou feel’st it cold. (Complete Sonnets and Poems 385)

The poem presents a clear-cut argument, which the poet presents by original images, organized in coherent conceits or image clusters: the years, referred to by the synecdoche of “winters,” are described by a military metaphor as they “besiege” the young man’s brow and get entrenched in his beauty’s field. The trenches may also refer to the furrows produced by a farmer’s plough in late autumn as well as the wrinkles on the aging addressee’s face. Lines 3 and 4 then provide a new image cluster: that of beauty being compared to the beautiful livery of a nobleman’s servant. First a besieged town or a piece of farmland, then a piece of clothing, the young man’s beauty next becomes treasure which is hidden away, rather ironically, given that it has vanished altogether. The solution is a fair child, who will carry on the bloodline and also be possessed of warm, youthful blood; various metonymical uses of blood interact in this image.

Apart from being so rich in imagery, the poem presents us with an intricate and unusual point of view. Not only does the poet enter into [→ 71] the mind of the addressee, he even enters the addressee’s mind as projected into the future, telling his readers what the young man’s appearance and thoughts might be like twenty or twenty-five years hence.¹⁶ The creation of such a point of view certainly testifies to the abilities of a dramatist used to fashioning the mental make-up of so many different characters.

Once we add the sonnet’s formal perfection, it becomes obvious that with sonnet 2 Shakespeare was at the height of his poetical powers, and there is no wonder that readers were so fascinated by it that they took manuscript copies and included it in anthologies. What, then, is wrong with sonnet 2? Why has it fallen out of favour with nineteenth-, twentieth-, and twenty-first century readers and anthologists, after having been so popular in the seventeenth and eighteenth centuries? To answer this question, I propose to look at two sonnets which are regularly found in anthologies and on reading lists, sonnets 30 and 116.

Let us begin with sonnet 30:

When to the sessions of sweet silent thought

I summon up remembrance of things past,

I sigh the lack of many a thing I sought,

And with old woes new wail my dear time’s waste;

Then can I drown an eye (unused to flow)

For precious friends hid in death’s dateless night,

And weep afresh love’s long-since-cancelled woe,

And moan th’ expense of many a vanished sight;

Then can I grieve at grievances fore-gone,

And heavily from woe to woe tell o’er

The sad account of fore-bemoanèd moan,

Which I new pay as if not paid before.

But if the while I think on thee (dear friend)

All losses are restored, and sorrows end. (Complete Sonnets and Poems 441)

The speaker indulges in memories about friends who have died and love affairs which have ended unhappily, before complimenting his friend on compensating for the loss of previous friends and lovers. The subjectivity of the speaker’s feelings is emphasized through the initial “when” clause. It is in “sessions of sweet silent thought” that he regrets [→ 72] the loss of friends departed and takes to crying while his eyes are “unused to flow” at other times.

Comparing the two sonnets, we notice that the speaker of sonnet 2 does not refer to himself at all. All of his messages are second-person messages. The speaker does not just give advice to his addressee but even takes possession of his mind. The speaker of sonnet 30, by contrast, provides a chain of first-person messages, indulging in that kind of self-pity which is also found later, for example, in Milton’s “When I consider how my light is spent” (Milton 83-84), Keats’s “Bright star! would I were steadfast as thou art” (Keats 452), and Tennyson’s “But the tender grace of a day that is dead / Will never come back to me” (from “Break, break, break,” Tennyson 244). Is it, we may ask, that anthologists prefer poems containing first-person messages to those which contain addresses directed at another individual? May this preference perhaps be informed by poetic conventions which post-date Shakespeare, so that sonnet 30 to present-day readers represents what they are looking for when opening a book of poetry?

Let us proceed to sonnet 116. Discussing what love, i.e. true love, is, the sonnet can be categorized with the tradition of poems defining an abstract concept. The speaker uses images of marriage, navigation, and harvesting to convey his message that real or true love will last until doomsday. Comparing “116” to sonnet 2, we may notice that both poems convey a message, but that the message of sonnet 2 is much more complex and original. Is it that readers and anthologists prefer a commonplace idea to an original one? At any rate, the message is more abstract and does not involve an addressee. The speaker only briefly refers to himself in the final couplet.

It appears obvious that it is not for reasons of formal quality that sonnets 30 and 116 are preferred to sonnet 2. From the points of view of form and imagery, all of them reach the highest standards of poetic excellence. The three sonnets, however, seem to represent three different communicative modes. My suggestion that anthologists prefer sonnets involving first-person messages, or non-personal statements, to those which address a second person, is so far only a tentative one, based on [→ 73] a reading of three sonnets out of a corpus of 154. I therefore propose to classify all of the sonnets according to their modes of communication, making use of the model of language functions established by Roman Jakobson in his seminal essay, “Linguistics and Poetics” (1960). Jakobson defines the “constitutive factors in any speech event, in any act of verbal communication” (21) as follows:

The addresser sends a message to the addressee. To be operative the message requires a context referred to [...], graspable by the addressee, and either verbal or capable of being verbalized; a code [...] common to the addresser and addressee [...]; and finally, a contact, a physical channel and psychological connection between the addresser and addressee, enabling both of them to enter and stay in communication. (21)

As Jakobson explains, “each of these six factors determines a different function of language”. In any message, these functions are placed “in a different hierarchical order”, and “the verbal structure of a message depends primarily on the predominant function” (22). Jakobson calls the function focused on the context “referential,” on the addresser “emotive” (22), on the addressee “conative” (23), on the contact “phatic” (24), on the code “metalingual” (25), and on “the message as such” “poetic” (25).¹⁷ If we apply this model to the three sonnets studied above, it is obvious that in sonnet 2 the conative function predominates, as opposed to the emotive function in sonnet 30, and the referential function in sonnet 116. Other sonnets (for example, 18 and 130) are primarily devoted to the art of writing sonnets and thus fulfil a metalingual function, according to Jakobson’s terminology; they could be called “poetological” or self-referential.

While not all of the sonnets lend themselves easily to this kind of classification, it is often possible to determine a predominant function on the basis of the quantity of pronouncements about either the speaker, the addressee, or a non-personal referent. In sonnet 10, for example, the addressee is clearly focused on, even though the speaker also briefly refers to himself in lines 9 and 13. All of the statements have an appellative or “conative” character. Sonnet 12, by contrast, can be classified [→ 74] as emotive, as lines 1 to 8 exclusively refer to the speaker’s perceptions and thoughts. It is only in lines 9 and 10 that the addressee comes into play. A similar assessment can be made with regard to sonnet 15, which in lines 1 to 8 again refers to perceptions and thoughts of the speaker and only turns to the addressee from line 9 onwards. Sonnet 32 raises the issue of what the young man should do with the speaker’s poetry after the latter’s decease. While the sonnet contains both a self-referential (“my poetry is no good”) and a conative message (“please preserve my poetry as a token of my love for you”), the conative function appears to be predominant. In sonnet 41 the conative function predominates as well, as lines 1 to 6 and 9 to 12 contain statements about the addressee, the young man, even though a woman, possibly the dark lady,¹⁸ is referred to in lines 7, 8 and 13. In sonnet 73 all of the pronouncements of lines 1 to 12 concern the speaker’s aging process, so that in spite of the fact that the statements are addressed to another person in lines 1, 5, and 9, and in spite of the two last lines which refer to this addressee, the sonnet is predominantly emotive.¹⁹

According to my reading, first-person, or emotive, messages seem to be predominant in 26 sonnets, while 33 sonnets (including most of the procreation sonnets) clearly privilege second-person, conative, statements (see Appendix II). Eight sonnets refer to either the young man or the dark lady in the third person; they can thus be called referential. In two sonnets (50 and 51) the speaker’s horse is the character whose train of mind the speaker tries to delineate. Another type of referential predominance is found in ten sonnets which discuss a non-personal referent, such as true love in sonnet 116 and “lust in action” in sonnet 129. Sonnets 153 and 154 tell a mythological story and thus form a third type of referentiality. Fourteen sonnets are self-referential. As nine sonnets refer to the speaker’s entanglement with two other persons (the young man and the dark lady, and the young man and the rival poet) and cannot easily be classified as either predominantly conative or referential, I propose to relegate them to a category of their own.

We are then left with a large group of sonnets which resist a classification, as first- and second-person messages are set side by side: the [→ 75] emotive and conative functions are clearly of equal importance, and subtly intertwined with one another. As an example, I should like to quote sonnet 61:

Is it thy will thy image should keep open

My heavy eyelids to the weary night?

Dost thou desire my slumbers should be broken,

While shadows like to thee do mock my sight?

Is it thy spirit that thou send’st from thee

So far from home into my deeds to pry,

To find out shames and idle hours in me,

The scope and tenure of thy jealousy?

O no, thy love, though much, is not so great:

It is my love that keeps mine eye awake,

Mine own true love that doth my rest defeat,

To play the watchman ever for thy sake.

For thee watch I, whilst thou dost wake elsewhere,

From me far off, with others all too near. (Complete Sonnets and Poems 503)

Throughout this sonnet thou- and I- pronouncements are subtly interwoven. “Thy will” (line 1) relates to “my heavy eyelids,” the young man’s desire may break the speaker’s slumbers (line 3). His spirit may pry into the speaker’s deeds (lines 5-6), to find out instances of idleness in him (line 7), to feed his, the young man’s, jealousy (line 8). The dialogical quality of the sonnet is enhanced when the speaker’s suppositions are refuted in lines 9 to 14, with “thy love” (line 9) being replaced by “my love” (line 10). Both the speaker and the young man are awake at night-time, but while the speaker is “watching” anxiously, the young man is “waking,” i.e. engaged in some kind of party.²⁰ The supposition of the young man’s jealousy in line 8 is replaced by a hint at the speaker’s jealousy in the last line of the sonnet. As Helen Vendler notes, we are then in a position to construe “the octave [lines 1-8] as a projection of the speaker’s own agony” (289).²¹

While the young man is separated from the speaker (we may be allowed to say: the poet) by physical distance, he is inextricably linked to him by means of language, metre, and rhyme. We may even say that rhetoric and poetry serve as means of sublimating erotic desire. While [→ 76] the poet’s love cannot find fulfilment in a physical way, he manages to intertwine himself with the young man in subtle language games, to become one with him in the construction of a sonnet.²²

In the body of Shakespeare’s sonnets this interweaving of first- and second person messages recurs again and again; in fact, it is 50 sonnets altogether which, I think, can be classified thus. If we were looking not for the “predominance” but just the prevalence of the emotive and conative functions, the count would be even higher, as the thou-perspective is rarely completely absent from the poems predominantly emotive. According to Giorgio Melchiori, the Shakespearean sequence contains 21 “I-less sonnets” (19) and 33 “non-You sonnets” (28-29).²³ While fifteen of the sonnets without a first-person pronoun also belong to my list of conative sonnets, only eight of the 26 emotive sonnets do not contain a second-person pronoun.²⁴ In eighteen of them, the poet’s concentration on himself is supplemented by an address to the young man or dark lady.

This rhetorical strategy may to some extent have been inspired by Sidney’s Astrophil and Stella cycle, which reports a kind of dialogue Astrophil has with Stella. In most of the sonnets, however, Astrophil refers to Stella in the third person. By my count, this applies to 66 sonnets out of the corpus of 108, while Stella is addressed in 21 sonnets. Astrophil is telling his readers the story of his unrequited desire for, and Platonic love affair with, this courtly lady. By contrast, the rhetoric interweaving described with regard to sonnet 61 seems to be a feature which is unique to Shakespeare.²⁵ We may well assume that it is Shakespeare’s competence as a dramatist which makes him create these I-and-thou exchanges, and they certainly show the most original side of Shakespeare the poet. As Sandra L. Bermann points out, “Shakespeare’s portrait of an ‘I’ and a ‘thou’ distinguishes his sonnets almost as radically from his English predecessors as from Petrarch himself” (73).

This subjective assessment can be supplemented by some hard statistical data collected by Melchiori (see 198): computerized concordances of the sonnet sequences by Sidney, Daniel, Drayton, Spenser, and [→ 77] Shakespeare show that 14.17% of all the words of Shakespeare’s sonnets are personal pronominal forms (like I, me, myself, my, mine, etc.). The proportion of pronominal forms used by Shakespeare is higher than that of the other four sequences but not strikingly so; the average being 13.2%. Even more significant are the proportions of first-person, second-person, and third-person pronouns: while in the five sequences the average proportion of first-person pronouns is 43.8%, it is 40.3% in the case of Shakespeare. The average proportion of second-person pronouns is 25.6%; in Shakespeare’s sonnets it is 37.2%. Concerning third-person pronouns the average figure is 30.6%, and 22.5% in Shakespeare. There are two conclusions we can draw from these figures: first, that intersubjective communication is central to Elizabethan sonneteers in general, and, second, that Shakespeare accords particular prominence to second-person addresses and is less interested in third-person propositions than other Elizabethan poets.²⁶ We can add that, while the addressees of Sidney’s sonnets include Cupid, the moon, Morpheus, hope, a kiss, a sparrow, absence etc., 120 of Shakespeare’s 128 sonnets which contain a second-person pronoun are addressed to the young man or the dark lady.²⁷

4. Canonical Preferences

To conclude our investigation, it now remains to correlate our categorization of the 154 sonnets with the frequency list. How many sonnets from each of our categories did our anthologists choose?

The answers are as follows: if we proceed from the list of 45 sonnets which were anthologized five times or more, we see that thirteen of them focus on first-person messages, seven on second-person messages, another ten belong to those which express the mutuality described. Two poems feature third-person messages, two are about triangular relationships, four are self-referential, and seven sonnets (out of a total of ten) express non-personal reasoning. We can notice a strong bias in favour of first-person-message and non-personal poems.

 [→ 78] The figures are even more striking if we proceed to the shorter list of 27 sonnets which were chosen to represent Shakespeare in ten or more anthologies: seven of the 27 sonnets focus on first-person messages, two refer to the young man or the dark lady in the third person, four of the sonnets are self-referential and seven express non-personal reasoning. By contrast, there are only two second-person sonnets and five expressing I-and-thou mutuality.

What are the reasons for this universal preference for sonnets which focus on first-person and non-personal messages? The roots of this practice may lie with Palgrave’s phenomenally successful Golden Treasury.²⁸ Out of the twenty sonnets chosen by Palgrave, six focus on first-person and two on second-person messages. The I-and-thou mutuality is found in four of the poems anthologized, two are self-referential, and six convey non-personal wisdom. Three of the eighteen sonnets included by Kearsley in 1783, by contrast, belong to the first-person group, three to the second-person, three to the I-and-thou mutuality, while five are non-personal.²⁹ We see that the two anthologies share a bias in favour of non-personal sonnets, while Palgrave also privileges those in which the first person predominates, in keeping with the penchant for Romanticism which Palgrave’s collection exhibits. Romantic poets, including Wordsworth, Coleridge, Byron, Shelley, and Keats, are lavishly represented. These Romantic poets obviously corresponded to Victorian notions of what poetry should be like; and poems in which a poet expresses his “Weltschmerz,” his suffering caused by the ways of the world and the human condition, fit in well with this Romantic discourse.³⁰ As Emrys Jones notes in the introduction to his New Oxford Book of Sixteenth-Century Verse, Palgrave’s anthology “had the influence it did because he was fully in sympathy with the main direction of Romantic literary theory” (xxvi).³¹

As stated above, most of the twenty sonnets chosen by Palgrave still constitute the staple of contemporary anthologies.³² It would be wrong, however, to lay the blame for the canon’s bias on Palgrave alone. The preference for first-person sonnets and sonnets containing [→ 79] non-personal reasoning became even more pronounced in some of the later collections. Robert Whiteford, an American anthologist, included four Shakespearean sonnets in his 1903 Anthology of English Poetry, three of which contain first-person messages, while one, sonnet 116, falls into the category of non-personal reasoning. In the German anthology published in 1910 by Westermann-Verlag (edited by Max Förster), three of the five sonnets chosen focus on first-person messages (30, 33, 73), one is self-referential (18), and one is non-personal (116).

In other early twentieth-century anthologies, this imbalance is not that obvious. Arthur Quiller-Couch, in particular, presented a rather original collection in his Oxford Book of English Verse, first published in 1900. Out of the twenty sonnets chosen, five are first-person-sonnets, three focus on the second person, six express I-and-thou mutuality, two are self-referential, and four contain abstract reasoning. Similarly, E. K. Chambers’s collection of “Fifty Sonnets” (676), contains twelve first-person, nine second-person, and fourteen I-and-thou sonnets, as well as three third-person, six self-referential, and six sonnets containing abstract reasoning.

The basic tendency in favour of first-person sonnets as well as sonnets containing self-referential statements and non-personal reasoning was reinforced, however, in the Norton Anthology, the later Oxford anthologies (Hadow, Hollander/Kermode, Leonard, Peacock, and Ricks), and Blaisdell’s collection of Elizabethan Poetry (2005), and transcended national boundaries: in anthologies published for study purposes in Germany (e.g. Meller/Sühnel), Poland (Mazur/Bela), and India (Chaudhuri), the biases mentioned are much in evidence. Some recent collections discard first-person sonnets as well and concentrate on self-referential and non-personal sonnets, rendering their choices even less representative, for example, Ricks, and Löffler/Späth. The Bedford Anthology of World Literature (2004) includes four Shakespearean sonnets, two of them self-referential (18, 130), and two non-personal (116, 129). While this publication venture has been hailed as ground-breaking in [→ 80] its extensive inclusion of non-European writing, the choice of Shakespeare sonnets selected to represent European literature cannot be considered either representative or revolutionary.

Looking at the “Norton canon” of the 24 sonnets shared by Norton with two other college anthologies, Longman and Broadview, we cannot observe any significant change with regard to the communicative categories researched: in six sonnets, the first-person point of view is predominant, four privilege the second person, three express I-and-thou mutuality, four are self-referential, and another four non-personal. Two sonnets concern triangular relationships, and one is a third-person sonnet. While some of the sonnets from the “Palgrave canon” have been replaced, the biases analysed remain intact.

If we discard the eighteenth-, nineteenth-, and early twentieth-century anthologies consulted and focus on the anthologies from the second half of the twentieth century onwards, the number of I-and-thou sonnets found in at least five anthologies is reduced from ten to four, and not a single one of the second-person or I-and-thou poems scores ten or more entries, i.e. 83 of the 154 Shakespearean sonnets are not represented in the group of sixteen sonnets printed most often. Looking at the three sonnets which found favour with Victorian anthologists but were dropped later, we may also note that two (57, 109) express I-and-thou mutuality while one (54) focuses on a second-person address.

Recent non-scholarly publications, though, have to some extent widened, or re-opened the canon. The Flame Tree collection of 2014 includes six sonnets not anthologized elsewhere, three of which belong to the I-and-thou group. The internet-based “Poetry Foundation” also includes some hitherto uncanonized sonnets from the second-person and the I-and-thou categories. Both collections contain sonnets which had been anthologized in Victorian and Edwardian anthologies but had been neglected since, e.g. 53, 54, 57, 98, 111, 148. It may be of some significance that these two publications address general readers, not scholars or students—who are obviously considered to be fixated to established traditions of scholarship.

 [→ 81]

5. Conclusions: Textual Reasons for Canonicity

By way of conclusion we can first state that, while the habit of publishing selections from Shakespeare’s 154 sonnets started with Kearsley’s Beauties of Shakespeare in 1783, it was with Palgrave’s Golden Treasury that a certain canon of sonnets emerged which in the nineteenth and twentieth centuries would be anthologized over and over again. This “Palgrave canon” would be slightly modified in the second half of the twentieth century, with sonnets 20 and 130 entering the canon, while a few others were dropped. A certain shift took place with the publication of college anthologies of English Literature like the Norton Anthology. The “Norton canon,” however, retains a core group of around ten sonnets which had already been part of the Golden Treasury selection. New departures are taken by certain non-scholarly anthologies.

Secondly, our investigation has shown that, from Palgrave onwards, anthologies display a marked preference for three categories of sonnets: those in which the speaker expresses his feelings in first-person statements, those which discuss certain issues from a non-personal point of view, and those which focus on the art of sonnet-writing itself. These categories correspond to the “emotive,” the “referential,” and the “metalingual” functions of language in the communicative model established by Roman Jakobson. “Conative,” or second-person sonnets, as well as sonnets which set emotive and conative pronouncements side by side (the two groups which together form the bulk of the Shakespearean corpus), however, have largely been neglected.

With regard to the preference given to first-person and non-personal sonnets, the aesthetic values of Romanticism which informed Palgrave’s selection apparently continue to set expectations as to what poetical excellence amounts to.³³ There may, however, be additional, and more specific, reasons to account for this preference: lovers of Shakespeare’s plays have turned to the sonnets to find out about the dramatist’s inner self and therefore focused on first-person sonnets,³⁴ overlooking the fact that Shakespeare’s genius is not least due to his ability to bypass his own inner self to enter into the minds of his characters. [→ 82] Considering the whole corpus of sonnets we may also argue that Shakespeare’s inner self was inextricably bound up with that of other persons, and that, like the plays, the sonnets testify to what Keats called Shakespeare’s “negative capability” (Houghton 62).

Readers have also searched through the sonnets for spiritual guidance. According to a traditional assumption, a great poet such as Shakespeare must have been possessed with incomparable wisdom, inferior only to that of the Bible.³⁵ It was the non-personal sonnets which most clearly fulfilled that demand.³⁶ Sonnet 116 was read and studied as conveying the truth about true love and offered to young people as a help to get a proper direction in life.³⁷ Such a reading could be compared to the mis-reading of the famous speech delivered by Jacques in As You Like It: “All the world’s a stage / And all the men and women merely players [...]” (2.7.139-40, Riverside ed.) etc. This speech was learned by heart by generations of school children as conveying Shakespeare’s wisdom; its dramatic context, which in a way refutes the pessimistic message conveyed by this speech, was not taken into account.

The preference given to self-referential sonnets also requires an explanation. When I studied English in the 1980s, sonnets 18 (“Shall I compare thee to a summer’s day”) and 130 (“My mistress’ eyes are nothing like the sun”) were taught as being “anti-Petrarchist,” and Shakespeare was praised for boldly and provocatively breaking away from Petrarchan conventions; since then I have repeatedly come across this interpretation. In view of the sonneteering output of other Elizabethan poets, however, I would like to argue that as an iconoclast Shakespeare has been overrated, as witty departures from Petrarchan conventions had become a staple of Elizabethan sonnets even before Shakespeare; Sir Philip Sidney and Michael Drayton, for example, could also be called anti-Petrarchists, for similar reasons (see Bermann 86). The I-and-thou poems discussed certainly constitute a more signi-ficant departure from Petrarca, most of whose sonnets to Laura are first-person messages, detailing the poet’s own woes and frustrations. However we account for the anthologies’ preferences, it is obvious that [→ 83] they all convey a very biased idea of what the sonnets are about. Shakespeare was made to conform to expectations about great poetry, and this way reduced to a size which readers from the Victorian Age onwards could, and can, handle.³⁸

What are the results of our investigation with regard to the more general questions about canonicity raised initially? One result is that there are indeed textual features which lead to a sonnet’s adoption into an anthology and which initiate a tradition of canonization. These features, however, obviously do not indicate any timeless aesthetic quality. Changes in the canon are rather due to the impact of cultural movements like Romanticism or a more recent interest in iconoclasm and non-heteronormative sexuality.³⁹

Some of the editors of modern anthologies explicitly refer to changing times as the reason for altering the selection of texts included. As editor of the Oxford Anthology of English Poetry, John Wain speaks of the “gradual obsolescence of [...] English Verse, edited by W. Peacock” (which had also been published by Oxford University Press) and claims that his “new collection” is meant to “serve the needs of a different time” (xix). Meyer Abrams, who, in the six edition, repudiates the charge that the editors of the Norton Anthology “reproduce, or even help establish, the traditional ‘canon’ of English literature,” asserts that the selection follows the requirements of schools and universities: “Some texts, which our canvass of teachers showed to be assigned infrequently or not at all, have been replaced by others which were more in demand” (xxx). Emrys Jones, the editor of the New Oxford Book of Sixteenth Century Verse, states that Edmund Chambers’s, his predecessor’s, volume “has receded further into the past” and “become more obviously not just a product but also an expression of its own time” (xxv). With regard to Shakespeare’s sonnets, however, it is hard to see in what way Jones’s selection of 43 sonnets (which to a large extent follows Chambers’s choice) improves upon Chambers’s 50: if we take the categories established in this contribution into account, we see that Chambers’s selection is clearly more balanced and representative than more recent anthologies (with the possible exception of Jones’s own).

[→ 84] We can conclude that a reexamination of a received canon might alert us to aspects overlooked by previous pedagogues and anthologists. Texts which were discarded or forgotten could prove to be more representative of their author and period than canonized works. In the case of Shakespeare’s sonnets a reexamination of the established canon has made us aware of the technique of mingling emotive and conative messages, which informs a considerable part of the sequence. Shakespeare’s sonnets thus provide another example of the truism that it is often the uncanonized texts which turn out to be the most interesting ones, as they correspond to our previous expectations least, and may thus teach us most about cultural history and—possibly—the human condition.

[→ 89]

Appendix II: Statistical Survey of Sonnet Categories

Sonnets anthologized five times or more:

1, 12, 15, 18, 19, 20, 25, 27, 29, 30, 32, 33, 35, 52, 53, 54, 55, 57, 60, 64, 65, 66, 71, 73, 86, 87, 90, 94, 97, 98, 104, 106, 107, 109, 110, 111, 116, 121, 128, 129, 130, 138, 144, 146, 147 (45 sonnets; 27 sonnets ten times or more, in bold)

first-person messages:

12, 14, 15, 25, 28, 29, 30, 33, 44, 47, 62, 66, 73, 97, 102, 110, 111, 113, 115, 118, 119, 121, 124, 137, 147, 148 (26 sonnets)

anthologized five times or more: 12, 15, 25, 29, 30, 33, 66, 73, 97, 110, 111, 121, 147 (13 sonnets)

second-person messages:

1, 2, 3, 4, 6, 7, 8, 9, 10, 11, 13, 16, 20, 32, 35, 40, 41, 48, 53, 54, 69, 70, 77, 82, 84, 93, 95, 96, 104, 126, 131, 132, 139 (33 sonnets)

anthologized 5 times or more: 1, 20, 32, 35, 53, 54, 104 (7 sonnets)

I-and-thou mutuality:

22, 24, 26, 27, 31, 34, 36, 37, 39, 43, 45, 46, 49, 52, 57, 58, 61, 71, 72, 74, 75, 78, 81, 87, 88, 89, 90, 91, 92, 98, 99, 103, 107, 108, 109, 112, 114, 117, 120, 122, 125, 128, 136, 140, 141, 142, 149, 150, 151, 152 (50 sonnets)

anthologized 5 times or more: 27, 52, 57, 71, 87, 90, 98, 107, 109, 128 (10 sonnets)

third-person messages:

19, 63, 67, 68, 105 (young man); 127, 138, 145 (dark lady/ Anne Hathaway), 50, 51 (horse) (10 sonnets)

anthologized 5 times or more: 19, 138 (2 sonnets)

self-referential content:

17, 18, 21, 23, 38, 55, 59, 76, 83, 85, 100, 101, 106, 130 (14 sonnets)

anthologized 5 times or more: 18, 55, 106, 130 (4 sonnets)

triangular relationships:

42, 79, 80, 86, 133, 134, 135, 143, 144 (9 sonnets)

anthologized 5 times or more: 86, 144 (2 sonnets)

non-personal reasoning:

5, 56, 60, 64, 65, 94, 116, 123, 129, 146 (10 sonnets)

anthologized 5 times or more: 60, 64, 65, 94, 116, 129, 146 (7 sonnets)

mythological story:

153, 154 (2 sonnets)