Beyond Close Reading: My Digital Humanities Lab Experience
Introduction
Our Digital Humanities lab sessions were conducted by Professor Dilip Barad, and he himself gave us these tasks on Google Classroom. Over the course of the lab, sir gave us four separate tasks. The first was to read his own blog post titled "What if Machines Write Poems?" The second was an NPR quiz where we had to guess whether a poem was written by a human or a computer. The third was CLiC, a tool that lets you search for patterns of words inside novels such as Great Expectations. The fourth was Voyant Tools, which turns a play like Macbeth into graphs, word clouds, and maps.
Before this, I always thought literary study meant one thing: sitting quietly with a book and reading it slowly, line by line, page by page. Sir's four tasks showed me a different side of the same subject. They did not replace close reading. They added a new way of looking at texts, using data and computer tools alongside the normal way of reading a novel or a play.
I am writing this blog to explain what each task was about, what exactly I did while completing it, and what I understood from it by the end of the lab session. I have tried to keep the explanation simple and close to what actually happened in class, rather than adding too much personal commentary, since the value of this exercise lies in the tasks themselves and in what they reveal about literature and technology.
Task 1: What If Machines Write Poems?
For the first task, sir shared his own blog post titled "What if Machines Write Poems?" which he had published on his blog on World Poetry Day. The post asks a question that sounds simple on the surface but is actually quite difficult to answer properly: what if machines can write poems that are as good as, or even better than, poems written by humans?
Read the full post:
https://blog.dilipbarad.com/2017/03/what-if-machines-write-poems.html
In the post, sir shares a video by Oscar Schwartz, a writer and researcher who has studied whether readers can tell computer poems apart from human poems. Sir also gives links to online tools where anyone can generate a poem by typing just a single line, after which a computer completes the rest of the poem automatically. Sir explains that this kind of poetry, written with the help of an algorithm trained on large amounts of existing poetry, is generally called generative literature or computer poetry.
Video: Can a Computer Write Poetry?
Reading this blog post before the lab session made me think about a question I had never seriously asked myself before. Does a poem need a real human feeling behind it in order to count as a poem at all? Or is a poem simply defined by its final words on the page, regardless of who or what produced them? Machines can now study patterns of rhyme, rhythm, imagery, and word choice from thousands of existing poems, and then use those patterns to generate new lines that follow the same structure. The final output can look and sound very much like a genuine poem written by a person. But it is built entirely from statistical patterns learned from other texts, not from personal memory, lived experience, or emotion in the way we usually understand these terms in relation to human poets.
Sir's blog post also raises a related point about how literature classrooms have traditionally treated the author as central to the meaning of a poem, and how machine-written poetry disturbs that assumption in an interesting way, since there is no single author with a personal history standing behind the words. This first task was mainly meant to open our minds to this problem before we tried to actually test our own ability to identify computer poems, which is exactly what the second task, the NPR quiz, allowed us to do.
Task 2: The NPR "Human or Machine" Poetry Quiz
For the second task, sir linked us to NPR's interactive quiz titled "Human Or Machine: Can You Tell Who Wrote These Poems?", the same quiz that is also linked from within his own blog post. The quiz presents six poems, one at a time, and asks the reader to guess whether each poem was written by a human being or generated by a computer program.
Quiz Example: "Paint the Sun" - Human or Machine?
The Challenge: Distinguishing Human from Machine Poetry
I read each of the six poems carefully before choosing my answer, rather than guessing quickly. Some of the poems were short and rhymed very neatly, such as the one that reads, "The sun's gone dim, and the moon's turned black, for I loved him, and he didn't love back." Others were longer and more descriptive in nature, such as the poem that mentions "a torpid badger sleeps in their fantasies," and another that reflects on walking to a street corner and wondering what would happen if one's own reflection turned the opposite way.
At the end of the quiz, the website reveals the correct answer for every poem, along with the name of the person who wrote it, or the name of the computer program and the researchers who built it. I got five out of six answers correct. The one poem I got wrong was the short rhyming poem about heartbreak, which I guessed was written by a computer because it felt almost too neat and too simple to be human. It actually turned out to be written by a human poet named Kurtis Hessel. Among the machine-written poems in the quiz, one was created by a program called the "Pythonic Poet," built by a team of researchers at the University of California, Berkeley. Another machine-written poem in the quiz was generated by a program developed by researchers at the University of Southern California's Information Sciences Institute.
My Quiz Performance: Testing Human vs Machine Detection
📊 MY QUIZ RESULT
5/6
Correct Identifications
The one poem I got wrong was written by Kurtis Hessel — I assumed its neat rhyme scheme indicated machine authorship. This mistake taught me that simplicity does not equal artificiality.
This task taught me something I did not fully expect before starting it. It is genuinely becoming difficult to guess who wrote a poem simply by reading it closely. A short, simple, neatly rhymed poem is not automatically a computer poem, and a long, descriptive, image-heavy poem is not automatically written by a human being. My own wrong guess proved exactly this point about myself. The quiz page also includes a quote from Dan Rockmore, a professor and director of the Neukom Institute for Computational Science at Dartmouth College, who says that this kind of exercise tells us more about what it means to be human than about what it means to be a machine trying to sound human. I found this a genuinely useful way to think about the entire activity, since it shifts the focus away from testing the machine and toward testing our own assumptions as readers.
Task 3: Exploring the CLiC Tool with Dickens
For the third task, sir asked us to use CLiC, which stands for Corpus Linguistics in Context, a free online tool developed jointly by the University of Birmingham and the University of Nottingham. CLiC is built around what is called corpus stylistics, a method that uses computer-assisted searching to study literary texts and to notice patterns that are difficult to catch through ordinary reading alone. The main collection of texts built into CLiC is a set of novels by Charles Dickens, along with some children's literature and other smaller text collections available on the same platform.
CLiC Tool: Corpus Linguistics in Context
Access CLiC:
https://clic.bham.ac.uk/
Using CLiC's Concordance search function, I searched for a set of words connected to gratitude and guilt inside Great Expectations: grateful, gracious, thankful, remorse, regret, and guilt. I also searched separately for the exact phrase "brought up by hand," a line connected to how Mrs. Joe describes raising young Pip. CLiC lists every single place where a searched word or phrase appears across the whole novel, along with a few words of surrounding context on either side, so that a reader can see the full pattern at a glance rather than having to remember where each instance occurred while reading the book from start to finish.
This search showed me clearly that "brought up by hand" is not just a single memorable line that appears once early in the novel. It actually returns more than once across the story, working as a kind of quiet reminder of how Pip was raised and how that upbringing continues to shape the way other characters treat him. The words connected to guilt and remorse also cluster tightly around a few specific moments in the novel, almost all of them tied to Pip reflecting on how he has behaved toward Joe and Biddy after coming into money and moving to London. Reading the novel from beginning to end, I had only vaguely sensed that this guilt builds up slowly over many chapters. Seeing every instance gathered together in one list, through CLiC, made this slow build far easier to notice and describe clearly in writing.
Tracking Patterns: Guilt and Gratitude Across Great Expectations
CLiC also offers other functions besides Concordance, such as Clusters, which shows repeated groups of words appearing together, and Keywords, which compares the vocabulary of one text against another to see which words stand out as unusually frequent. For this particular task, we mainly relied on the Concordance function, since it was the most directly useful tool for tracking specific words and phrases across the full length of the novel.
Task 4: Visualising Macbeth with Voyant Tools
For the fourth and final task, sir introduced us to Voyant Tools, a free web-based platform for text analysis and visualisation. We uploaded the Folger Shakespeare Library edition of Macbeth into Voyant and explored six of its different modules: Cirrus, Trends, Contexts, Loom, Constellations, and DreamScape.
Cirrus
Cirrus: Word Cloud Showing Frequency in Macbeth
Cirrus displays a word cloud, where the size of each word on the screen depends directly on how often that word appears across the whole play. Along with "macbeth" itself and the line-numbering code "ftln," which repeats constantly throughout the Folger text file, words such as king, thane, hail, blood, and night appeared noticeably large in the cloud, matching the play's well-known themes of kingship, prophecy, and violence.
Trends
Trends: Tracking Word Frequency Across Segments
Trends plots how often a chosen word appears across different sequential sections of the play. When I checked the word "macbeth" itself across the ten sections that Voyant automatically created from the text, its frequency rose steadily through the early scenes, dipped slightly around the middle, and then spiked sharply near segment six, close to the point where Banquo is murdered and the banquet scene takes place, before falling again gradually toward the end of the play.
Contexts
Contexts functions like a concordance search built inside Voyant, working in a similar spirit to what we had already done using CLiC. Searching for the word "blood" here showed how the word returns again and again around Macbeth's own bloodstained hands, around Lady Macbeth's later obsession with washing imagined blood off hers, and around the general violence of battle described early in the play.
Loom
Loom: Word Frequency as Flowing Lines
Loom displays word frequency as a set of flowing lines moving across the whole text. This module took some time to properly understand at first glance. Once I understood how to read it, though, I could clearly see that only a small number of words rise noticeably above the rest of the lines, while most other words stay low and heavily overlap each other in the background, similar to a large crowd where only a few voices manage to rise above the general noise.
Constellations
Constellations: Network of Associated Words
Constellations shows which words tend to appear near each other most often, displaying them as a connected network rather than as a simple ranked list. The cluster built around lady, night, blood, love, and know matched almost exactly what literary critics have traditionally said about Lady Macbeth as a character, linking her closely to themes of guilt, sleeplessness, and blood imagery, all formed automatically from word patterns in the text itself.
DreamScape
DreamScape: Geographic Mapping of Place Names
DreamScape takes place names mentioned anywhere in the uploaded text and plots them directly onto a real, modern world map. Since Macbeth is set in eleventh-century Scotland and does not really engage with most modern countries in any meaningful way, this particular map did not always feel accurate or especially useful when applied to the play. Even so, it was an interesting way to see how a general-purpose visualisation tool behaves when applied to a text that was never originally organised around modern geography in the first place.
What I Learned
Each of the four tasks that sir gave us on Google Classroom showed a different side of Digital Humanities as a field of study. His own blog post made me seriously question, for what felt like the first time, whether a poem truly needs a human feeling behind it in order to count as a poem at all. The NPR quiz showed just how difficult it has genuinely become to tell machine writing apart from human writing simply through close reading, since my own confident guess turned out to be wrong. CLiC took a pattern I could only vaguely sense while reading Great Expectations and turned it into a clear, searchable, and citable list of evidence. Voyant Tools did something very similar for Macbeth, especially through the Constellations module, which grouped Lady Macbeth's name together with blood, night, and guilt entirely on its own, without any critic or teacher instructing it to do so beforehand.
None of these four tools are meant to replace ordinary reading, and I do not think sir intended them that way either. Close reading remains the method that gives a text its interpretation and its deeper meaning. These digital tools instead give evidence and visible patterns that can support, test, or sometimes even challenge that interpretation. Completing all four tasks together, one after another, helped me understand more clearly why Digital Humanities is becoming an increasingly important part of literary studies today, not as a replacement for the slow, careful reading we are trained to do, but as an additional method that works alongside it and strengthens it with actual evidence drawn directly from the text.
💡 KEY LEARNINGS FROM DIGITAL HUMANITIES LAB
Machine Poetry
Challenges authorship assumptions
Pattern Detection
CLiC reveals hidden structures
Data Visualization
Makes patterns visible
Complements Reading
Not a replacement
Conclusion
These four tasks, taken together, taught me that Digital Humanities does not replace close reading. It supports it with evidence. Sir's blog post raised the question of what counts as a poem, the NPR quiz tested my own ability to answer that question in practice, and CLiC and Voyant Tools showed me how patterns I could only sense while reading can be turned into clear, visible proof. Close reading still gives a text its meaning. These tools simply help me show that meaning more clearly.
📚 Works Cited
Barad, Dilip. "What If Machines Write Poems?" Dilip Barad, 21 Mar. 2017,
blog.dilipbarad.com/2017/03/what-if-machines-write-poems.html
"CLiC Dickens." CLiC, University of Birmingham and University of Nottingham,
clic.bham.ac.uk/
Palca, Joe. "Human Or Machine: Can You Tell Who Wrote These Poems?" NPR, 27 June 2016,
www.npr.org/sections/alltechconsidered/2016/06/27/480639265/human-or-machine-can-you-tell-who-wrote-these-poems
Voyant Tools, created by Stéfan Sinclair and Geoffrey Rockwell,
beta.voyant-tools.org/
This reflection explores how Digital Humanities tools — from machine-generated poetry to corpus linguistics platforms and text visualization software — complement traditional literary study by making textual patterns visible and testable.
No comments:
Post a Comment