Quantifying the Miltonic Sonnet

(This paper was presented at the University of British Columbia in a joint session of the Canadian Society for Renaissance Studies and the Canadian Society for Digital Humanities at Congress 2019. You can download the complete slideshow as a PDF, here.)

This paper began with a simple question: what is a Miltonic sonnet? What features distinguish John Milton’s sonnets from those by other poets?

Answering that seems straightforward; just read them. Milton only wrote 24 sonnets: 18 English, 6 Italian. You can read them in an hour, form an impression, and summarize them.

So I did.

The Oxford edition of his Major Works orders them chronologically.

During Milton’s Italian travels of the late 1620s, he wrote his first series. They’re the work of a young poet, about love and nightingales. Milton has clearly read his Petrarch, and is keen to imitate him; he tells his friend Charles Diodati that he has “fallen … for a foreign beauty,” but it feels like a Petrarchan pose (Sonnet 4). He even imitates Petrarch’s language: 6 of the 7 are in Italian, and one is a 15-line ‘canzone’ on the Petrarchan model.

Why Italian? Because “this is the language of which love is proud,” he writes (Canzone). “Love wakens on my quick tongue the strange flower of a foreign language” (Sonnet 3).

But I suspect this is Milton just returning to the sonnet’s source, ad fontes, to learn the form and its conventions: a 14-line lyric, traditionally on a subject of self-inquiry or personal feeling, usually in iambic pentameter. Originally the sonnet was divided into an octave and a sestet — until English writers in the Elizabethan sonnet craze (including Shakespeare) developed an alternate rhyme scheme: three quatrains and a couplet.

Mention sonnets, and Shakespeare is the author most people think of. But Milton’s sonnets are quite unlike Shakespeare’s: they’re not sequential, for one thing, and they’re resolutely Italian in form.

This poet, who dispraised the barbarous custom of rhyme to justify using blank verse for Paradise Lost, opted for an older, foreign model for this lyric form.

How else do Milton’s sonnets differ from Shakespeare’s? They have a far more devotional and spiritual tone. They praise women for biblical virtues, not for beauty; instead of a summer’s day, Milton says, “Shall I compare thee to Ruth, from the Old Testament?” (I’m paraphrasing.)

Milton’s sonnet also differ in subjects, beyond form. After these initial 7 Italian sonnets, the remaining 17 are about public or private occasions: deaths, dedications, critiques of his treatises, his blindness.

There are a few categories of Milton sonnets: those on specific occasions, like the Roundhead’s siege of London; and those praising specific people, like Henry Lawes or Margaret Ley or the unknown lady of Sonnet 9.

Two of them reflect on events in Milton’s own life.

In the early 1630s he regrets his unfulfilled ambitions now that “time … has stolen on his wing my three and twentieth year” (Sonnet 7). Milton attributes his actions, as he attributes public events, to “the will of heaven”; and when he praises others, it’s often for their heavenly virtues or their acceptance of heaven’s will.

In a later sonnet (16), written at midlife, he berates himself again for a lack of productivity, but resolves (in a memorable closing line) that “They also serve who stand and wait.”

We’ll return to that sonnet at the end of this paper.

I said at the beginning of this paper that I wanted to define a Milton sonnet. There’s a wider context for that problem: finding a satisfying, categorical definition of sonnets.

I now want to pose this question, and say how I plan to address it. Is the sonnet a form or a genre?

Our definitions always default to form — a 14-line lyric poem with a Petrarchan or Shakespearean rhyme scheme —because it’s the one constant through time. Since the Italians invented the sonnet in the 13th century, hundreds of poets have written many thousands of them in dozens of languages. What they all have in common is their form, with a few exceptions and experiments.

Another reason we default to formal definitions is the difficulty of a generic definition.

What would a generic definition of the sonnet even look like? Paul Oppenheimer tried, in 1989: a first-person reflection or “dialectical self-confrontation,” usually with a volta or turn: from problem to resolution, question to answer, mood A to mood B.

We think of Shakespeare: “And yet, by heaven, I think my love as rare, | As any she belied by false compare.” Or Donne: “And death shall be no more; Death, thou shalt die.” And so on.

Oppenheimer’s definition covers us the internal movement of the sonnet. But what about its subjects?

Thanks to Petrarch and his imitators, and his translations, sonnets were traditionally about love: unrequited, agonizing, beautiful love. When Thomas Wyatt wrote the first English sonnets, translating Petrarch, he set this tone.

To put it another way: Wyatt established a list of words that English sonnets contained. Each subsequent sonneteer expanded that list.

So just as Wyatt was the first to use words in a sonnet related to courtship, a later poet was the first to use words related to topics like marriage, or death, or (ultimately) ice cream.

To address the question of genre, I want to temporize that expanding list of words, and the topics they form.

Those digital humanists who’ve been wondering when this paper would move into familiar territory, the answer is now. Here’s my bridge between the Renaissance and the digital: my database of sonnets.

This is something that didn’t exist before. Sure, there were sonnets in text corpora — but undifferentiated, unregularized, without metadata.

My students and I started with a proof-of-concept database: just 445 sonnets transcribed from a single print anthology (Hirsch and Boland: 2008). When we added Project Gutenberg texts haphazardly, the results were unreliable: because the data was both incomplete and inconsistent. Spellings were all over the place, and some authors’ works were under-represented.

So we started again, and have since have transcribed 1,153 English-language sonnets from scholarly editions of every sonnet by more than a dozen poets of the 16th and 17th centuries. That is: everyone I’ve mentioned thus far, plus Surrey, Sidney, Spenser, Lodge, Daniel, Drayton, Constable, Drummond, Herbert, et al.

The database offers their sonnets in TEI-XML, or as JSON objects, and it maintains a Python class for data reuse via the RESTful API. The text-level metadata is light: just the author, the time period (in 50-year increments, for now), the copyright status, and some back-end details (user, date, &c).

I’ve deliberately used editions of these poets that critics will trust, for that reason — and because their modernized language limits the problem of variable orthography in (say) the EEBO-TCP transcriptions. That language also allows for more reliable lemmatization.

The database is set up to answer that research question I posed earlier: how do English sonnet subjects change through time? Who’s the first to write about God, or the Trinity, or Satan? What words are associated with them, and how do they change through time?

It’s time to show you some results. Let’s start by defining a Milton sonnet. Another way of asking this is: what distinguishes a Milton sonnet from a sonnet by someone else?

Here’s your answer. These are the words that Milton uses, that no other sonnet-writer does.

Notice how many of them are proper nouns. (All have been changed to lower-case to permit the comparisons.)

Right now there are a few duplicate copies of Milton’s sonnets in the database: he only uses terms like ‘cromwell’ or ‘tetrachordon’ once, I assure you.

But far more illuminating, frankly, are the words that other sonnet-writers use, that Milton’s 24 sonnets never use. Here they are.

Notice how Milton never uses words that seem associated with the syrupy love-sickness that we sometimes see in Elizabethan sonnets: sweet, beauty, desire, fair, pleasure. Nor with Petrarchan suffering: die, flame, pain, alas, ill, burn.

The recurrence of old-spelling words should be disregarded, because the process isn’t presently able to disambiguate between two spellings of the same word (‘love’ and ‘loue’), but I see that it ought to. The next step is to go through all the words that appear in lemma lists like this that aren’t in standard dictionaries and to modernize them.

Another immediately obvious problem arises, here. At present, none of Milton’s Italian sonnets are in the database, because they’re Italian — and translations would use lemmas chosen by translators. I don’t know, yet, how to address this problem. (Certainly we include contemporary translations: Wyatt began by translating Petrarch.)

Those non-English sonnets are not edge-cases: 1 in 4 Miltonic sonnets is in Italian (6 out of 24).

I deliberately began this talk with them. Partly because that’s how traditional criticism works: close reading, finding patterns, gathering evidence to build the big-picture understanding; it doesn’t begin with visualizations or lists like this.

The danger of digital capabilities is that they lead us toward quantified methods — rather than the texts themselves, and our linear readings of them, leading us there.

There is also the exploratory nature of these tools, not just the investigatory. There is what Stephen Ramsay called the hermeneutics of screwing around, seeing what the numbers say.

But I want to practice a form of traditional criticism that is informed by digital methods — not wield my hammer because every problem looks like a nail.

Partly I began with Milton’s Italian sonnets to show you that when we critics start with close-reading and then move outward to distant-reading methods, we can test whether it accounts for features that we know to be there.

And we can test the assumptions behind the data. In this case, my assumption that extra-English sonnets don’t count will be useful for a longitudinal history of the sonnet — but not for a narrower author-by-author understanding.


What can we glean from Milton’s 17 English-language sonnets? Let’s see how they compare to those of two contemporaries’ sonnets: two whose outputs are about the same in number, and comparable in devotional subjects.

John Donne has about twice the number of tokens in the database than Milton does: 6360 to Milton’s 3093.

George Herbert has about the same number of tokens in the database as Milton does: 2514 to (again) Milton’s 3093.

There are other ways I could investigate Milton’s understanding of a sonnet, beyond author-by-author comparisons. I could contrast his word-choices in the sonnets with his works in other forms. That would tell me if Milton saw sonnets as a subcategory of his writing.

In the comparisons I’ve shown you, I’ve deliberately used writers of whose sonnets you likely have a mental model. When I think of a Donne sonnet, or a Herbert sonnet, and I think of certain topics — or at least of a prevailing tone or mood. The same is true for Shakespeare, or for Sidney.

But I don’t have the same mental model of a Milton sonnet, probably because they’re obscured by longer works, and probably because they’re mostly occasional. And because he didn’t write a sequence, like others did: a mosaic pattern into which I can fit those individual tiles.

Can genre be localized in diction? Surely it’s not that simple. Consider this final example: Milton’s most famous sonnet of patience and fortitude. Consider the epigrammatic quality of the closing line, and its value of perseverance and service and humility, counterbalances the speaker’s initial midlife anxiety. It valourizes some of the exemplary values of Christ that the later Milton will praise in Paradise Lost: “the better fortitude | Of patience and heroic martyrdom.”

The speaker’s perturbations are reduced to a “murmur,” (l.9) as Patience replies by turning his word “serve” (ll.5 and 14) against him, as it were. She turns his his anxiety into her reassurance.

I show you this to conclude that genre is more than word choices. Yes, it’s the words and subjects that are proper to the genre. But it’s also how ideas change as the text unfolds, what relationship the speaker has to those ideas, what conclusions he draws about them, what his mental attitude is toward them, how he elaborates and refines and resolves them.

Spend your life standing and waiting, says Milton, and you’ll meet your stated intentions.

Leave a Reply