John Donne and the Sonnet Problem

What makes a sonnet? For most early modern examples, the answer is clear: a 14-line rhyming poem, its form either Shakespearean (three quatrains and a couplet) or Petrarchan (an octave and a sestet). There are exceptions to those formal rules, but most sonnets meet them.

Formal rules are the conventional answer. And that answer works for conventional sonnets, which are the vast majority of sonnets.

But if you enforce formal rules too rigorously, you encounter a few interesting problems. These are the problems that my project is investigating. Moments’ Monuments: The ACL Database is collecting as many sonnets as possible, so I can get a more definitive answer to this question: Is the sonnet a form or a genre? The trouble is, you need to decide first what qualifies as a sonnet.

My students in ENGL411 this term are entering 17th-century sonnets into the database, starting with a few major authors: Shakespeare, Milton, and Donne. I hoped that choosing them would be simple, but Donne is anything but simple.

Return to those formal rules. Say you begin with the premise that a sonnet must be 14 lines long. Must it really? Sonneto from the Italian means “a little song,” and there are plenty of so-called sonnets by Petrarch himself that break this so-called rule.

That’s fine, you say. Those Italians are always breaking rules. At least English poets are restrained by dignified rules. Surely no English sonnet is longer or shorter than 14 lines.

Wrong: Shakespeare’s own Sonnet 126 is deliberately truncated at 12 lines.

Well yes, you say, but that’s just one of Shakespeare’s 154. We can allow it as an aberration, without threatening the 14-line rule. It suits Sonnet 126’s theme of the young man’s truncated life — so Shakespeare’s being playful. Even innovative.

You get my point. A form-based definition is problematic, because of all the exceptions.

Donne’s Definition

Consider Donne.

Four years after his death, in 1635, a little book of his poetry was printed under the title Songs and Sonnets. With that title, surely it contains a sonnet or two. In fact, the book has no poems that we’d define formally as sonnets. There’s only one 14-line poem in this collection, “Witchcraft by a Picture,” and its form (two 7-line stanzas rhyming ABABCCC) is unlike any sonnet definition.

Another poem titled “Sonnet. The Token,” is 18 lines long: four quatrains, one more than the expected three, and a couplet. There’s no evidence that Donne gave the poem that title, but evidently someone in 1635 thought it was a sonnet.

Okay, then. But it fits the conventions of a sonnet in other ways: a speaker’s first-person address to his beloved, cataloguing the inadequate tokens she might send of her love, culminating in his turn (or volta) away from them in a conclusive couplet. Except for that extra quatrain, it reads like a textbook sonnet.

It might just be Donne, you say. He had some funny ideas about sonnets. Besides, his better-known sequence of 26 Holy Sonnets use the comfortable Shakespearean structure. Properly English, those. Others are just, well, experiments. Aberrations. Edge cases.

Verse Letters

Very well. Consider another category of the “verse letter.” A. J. Smith’s Penguin collection of Donne’s Complete English Poems gathers 37 of these scattered poems, of which 9 are 14 lines long. Yet not a single one of those 14-line poems adopts the sonnet’s formal definition.

Should they be defined as sonnets?

One (“To Mr B. B.”) is a 28-line poem comprised of two stanzas each with the Shakespearean sonnet rhyme scheme (ABAB|CDCD|EFEF|GG). Yes well, you say, that’s just two sonnets comprising a single poem. Shakespeare has plenty of similarly paired sonnets.

But if sonnets can be embedded in other, longer forms like the verse letter, do all of the embedded parts need to be of equal length? Consider one of Donne’s multiple poems addressed “To Mr R. W.,” whose 32 lines are split into two stanzas of 14 and 18 lines.

The first stanza is probably a sonnet, you say. But further inspection reveals that it has no quatrains; the whole thing (both stanzas, in fact) is couplets. Couplets don’t make a sonnet, so on second thought, no. This is not a sonnet.

Are you sure about that?

Yes, you say again. Not until later wild experimentalists like John Clare in the 19th century. Early sonnets definitively are not made of couplets.

Let’s work from that premise. If so, then the next poem (in Smith’s edtion) that Donne addressed “To Mr R. W.” is also not a sonnet because it’s 14 lines of couplets.

Nor is the second of two poems addressed “To Mr. I. L.” a sonnet, nor the third of four addressed “To Mr. T. W.” All couplets.

If couplets are out and quatrains are in, let’s split the difference and talk about tercets. Consider the fourth poem “To Mr. T. W.,” which rhymes AAA|BBB|CCC|DDD|EE. Sonnet?

No, say you. It still reads too much like an imploring letter.

Really? Then you be the one to tell Sidney that his entreaties in the Astrophel and Stella sequence aren’t sonnet-like enough. You’re wrong. Besides, think of all those dedicatory sonnets before Spenser’s Faerie Queene, rife with similar terms of fealty and favour.

Is the Sonnet a Form or a Genre?

The point is, as our form-based definition loses its validity, our only recourse is to a genre-based definition. Which has something to do with the register, the tone, the subject, the discourse, the diction and other qualitative features of these poems. Something like this: The sonnet is a first-person direct address, often to a beloved, often with a volta or turn.

The key word there is qualitative: those features are harder to quantify than numbers of lines, or rhyme schemes. Not impossible, just harder. That makes them more subject to interpretations and exceptions.

So what’s to be done with Donne?

I’m going to have students add all of his 14-line poems and stanzas into the database, so we can see how they compare quantitatively to his 26 Holy Sonnets. Compare verbally, that is: their diction and topics and sentiments and other quantifiable features that machines are capable of determining. But we human readers ought also to code them for qualitative features that machines can’t: like tone, or tropes, or topics.

Then we’ll be able more decisively to say which are sonnets, and which are too atypical — and thus define what ‘typical’ means, both for Donne and for his contemporaries.

Leave a Reply