Networks are everywhere in today’s digital world, so it is no surprise that in recent years there has been a so-called “network turn” within the academy.1 This is especially the case in early modern studies. Projects such as Six Degrees of Francis Bacon, Shakeosphere, and Networking Archives have all harnessed the power of network analysis to engage in literary and historical research. In this article, I use network analysis to enrich and decolonize approaches to reading Spenser. I reveal how, by utilizing network algorithms, it is possible to bring together authors and texts that share geographical space but inhabit different cultural spheres.
By network analysis I mean network science—the study of patterns of interconnectedness among a set of things or, as described by Mark E. Newman, the study of a set of relationships between different objects.2 For example, a network could be made up of people and map the interpersonal ties that unite them (familial, social, occupational etc.). These complex systems can be abstracted and represented as nodes (the objects in the network) and edges (the relationships that connect the nodes), resulting in visualizations. While these visualizations sometimes present information in a readable way, they are often too large and clustered to illustrate an argument. Instead, the critical power of network analysis is the use of network algorithms to analyze the underlying structure of the network and bring overlooked connections or entities to the fore.
Network algorithms and network visualizations can only represent the data that fuels them, so it is imperative to understand the provenance of data before you can engage with algorithmic processes. For this article, I draw on data curated and cleaned by a project I was project manager of, the MACMORRIS project at Maynooth University.3 This project developed a bilingual web application that explores the range and richness of cultural activity, across languages and ethnic groups, in Ireland, from 1541 to 1660. In the process of creating its database and network interface, we drew on three main sources: the Dictionary of Irish Biography (D.I.B.), the Bardic Poetry Database (B.P.D.), and the English Short Title Catalogue (E.S.T.C.).4 The data curated from these sources was abstracted into two types of networks: prosoprographical networks and bibliographical networks. The prosoprographical networks connect people to one another if a D.I.B. entry states that they were in some way acquainted. It also connects them if they are the author or patron of the same bardic poem and if they are the author, printer, bookseller, publisher, or dedicatee of a printed text. For example, Spenser is connected to Elizabeth I as he dedicated The Faerie Queene to her. Similarly, the bardic poet Eochaidh Ó hEódhusa (fl.1580–1607) is connected to the Ulster lord Cú Chonnacht Óg Mág Uidhir (1570–1608) as he wrote several poems for him. Equally, the bibliographical networks connect works to those who authored them, published, sold, or produced them, and those who received their dedications. In this instance, the bardic poem “A bhean fuair faill ar an bhfeart” [My lady who has found the tomb unattended] is connected to Nualaidh Ní Domhnaill (c.1575–c.1630) as the poem is written to her. Given Spenser’s crossover roles as a poet, colonial administrator, and settler, he is represented in both types of network, which makes him a good case study for considering how the algorithmic processes of network analysis can open up new ways of reading his texts comparatively.
First up is the prosoprographical network. To apply algorithmic processes to this network, I imported the data into Gephi, an open-source network analysis application. For this article, I applied four different algorithms to the data. The first algorithm I ran was the Louvain detection method, a community detection algorithm. This is an iterative process that analyzes the network grouping nodes based on the density of the edges between them and representing the groupings as a number.5 Once the algorithm has run, it is possible to color the nodes in the network based on their assigned community. In the resulting network visualization of this process (Fig. 1) Spenser is in the large pink cluster to the right of the network.
Fig. 1. Prosoprographical network drawn from MACMORRIS dataset, colored using the Louvain community algorithm.
So, who does this algorithm associate Spenser with? On the whole, Spenser’s community is fairly unsurprising in that it is made up of New English, colonial writers, and settlers who need no introduction—men like Lodowick Bryskett (1546–1612), Walter Ralegh (1554–1618), Barnaby Rich (1542–1617), Gervase Markham (ca.1566–1637), and Thomas Harriot (1560–1621). Few readers need data to confirm this intuition. However, as Matthew Lincoln argues, it is important to distinguish between what sounds reasonable in retrospect from what we knew before.6 In this instance, the above names do not need that distinction, yet there are interesting avenues to explore within this community. We might connect Spenser and Thomas Churchyard (1523–1604), rarely spoken of “in the same breath except in relation to the award—frustrated or otherwise—of a royal pension to an Elizabethan poet,” as Matthew Woodcock noted in a recent issue of this journal.7 Or Spenser and Walter Quin (1575–1650), an Irish poet and in Willy Maley’s words a “spectre in Spenser Studies.”8 Fellow Munster settler colonial poet Anne Southwell (1574-1636) is also in this community, a writer like Spenser who is rarely considered within the Irish contexts in which they produced their poetry. As the community algorithm picks up on Spenser’s core network it can be seen as a validation test. We want to see these connections in the network, otherwise we might not trust the algorithmic process. And yet a limitation to the Louvain model is that the algorithm can only map one community at a time and only assigns a node to one community, which does not align with the real world where a person would be a part of many different communities.
Despite this limitation, once we have this validation we can think about other ways to interrogate the network and employ different algorithmic processes. One of the most effective ways is to combine statistical algorithms to see what role a person plays in a network and thus to see which other people share a similar network profile. The three statistical algorithms I combined are: degree centrality, which measures how many connections an entity has; betweenness centrality, a measurement that takes into consideration that, for any two nodes in a network, there is a shortest path between them, and calculates how many of these shortest paths go through a given node; and eigenvector centrality, which measures how much influence a node has on other nodes and assigns a relative score from highest influence to lowest.9 Those who rank highly in all three measurements are the “hubs” or most influential figures of the network with a large number of connections. On the other hand, those who rank highly for betweenness, but less high for eigenvector, are “bridges” or “brokers,” those who connect discrete parts of a network. Having a similar network profile or playing a similar function in the network does not necessarily mean that one person is directly connected to another, nor does it mean that those that share a similar profile are in the same Louvain community. Therefore, the combination of these algorithms is a way to search the network for those who write in discrete linguistic traditions within the same cultural space.
In this network, Spenser is a broker: he ranks highly for degree, highly for betweenness, and less highly for eigenvector—that is, influence. When we look for people who have similar network profiles, we get five names (all of whom are writers), producing further potential for comparative study. Three were also grouped in the same Louvain community as Spenser: Churchyard, Markham, and Rich—again helping validate the algorithmic results. The fourth is James Shirley (1596–1666), which makes sense considering their respective positions in the canon and the importance of Ireland to their careers. However, a final figure offers the most potential for new avenues of comparative study—the bardic poet Tadhg Dall Ó hUiginn (c.1550–91).
Tadhg Dall Ó hUiginn was most likely born in the barony of Leyney (in modern County Sligo). He received his bardic training through at least two channels, within his family, and at the Uí hUiginn bardic school in Ceall Cluaine (in modern County Galway).10 With fifty-one poems attributed to him, Ó hUiginn has one of the largest extant oeuvres from the mid to late sixteenth century. He usually penned eulogies, petitions, and elegies to male patrons, including poems to Aodh Mag Uidhir [Hugh Maguire] (d.1600), and Aodh (mac Maghnuis) Ó Domhnaill [Hugh O’Donnell] (c.1520–90). He also has several extant house poems, aislings (vision poems), and satires. His two aislings, “Néall mná síthe sunn aréir” [There was a vision of a fairy woman here last night] and “An tusa an bhean do bhí sunn aréir tre amhra agum?” [Art thou the woman who was here last night with me in a vision?] offer new avenues for considering Irish fairylore in The Faerie Queene, especially considering Woodcock has already noted that Spenser likely came “in contact with the Gaelic tradition of fairy.”11 In both poems, Ó hUiginn discusses visions he had of a sidhe (fairy/mound-dweller), her beauty, and her connection to the Irish landscape. In the second poem, he uses fairy landscape imagery to describe the beauty of her body before enumerating those she was able to enchant to fight for her, including a certain King Arthur:
6. Cíche bláithgheala beaga ós leirg áluinn oighreada; cláir mhíne is caoimhfinne claidh, maoilinne sídhe a samhail. … 9. Nó an tú táinig eacht eile don Bhord Chruinn ar chéilidhe, a chiabh slim fionnfoltach úr, go Cing iongantach Artúr? [“Small, smooth, white breasts rising above a lovely, shining slope; gentle expanses, with borders most fair and delightful, they are to be likened to fairy knolls. … Or art thou she who came afore-time to visit the Round Table, thou head of smooth, fair, bright locks, to wondrous King Arthur?”] (A Bhfuil, poem 40)
This attempted enchantment of Arthurian knights is reminiscent of many elements in The Faerie Queene, but links most closely with Arthur’s vision of the Fairy Queen in Book I (I.IX.13-15) and to the adventures of Guyon, the knight of Temperance, in Book II.12
Both “Néall mná síthe sunn aréir” and Arthur’s vision (I.IX.13-15) draw on key fairylore conceits, while following a similar narrative structure. They open with a dream and a vision of a fairy woman appearing to a man to offer him her love. In Spenser’s poem this vision begins when a “royall Mayd/ Her daintie limbes full softly down did lay” (I.IX.13.7-8), while Ó hUiginn’s poem opens with a royal vision of a beautiful woman coming to the speaker in his sleep: “mairg fuair radharc an rínéill/… Ionmhuin taidhbhse táinig sonn/ aréir trem chodladh chugom [alas for him who beheld the royal vision … Dear the shape which came here to me last night in my slumber” (A Bhfuil, poem 39). Spenser’s only physical description of this fairy is her “daintie limbes,” while Ó hUiginn gives over several quatrains to praising the physical beauty of his fairy, from her eyes and brows to her lips and her breasts: “Béal tana nár mhillse mil/ ar ghné ógróis 'gon inghin/ … tolcha corra chíogh gcaoimhgheal,/ is díon orra d'órchraoibhreadh [ Slender lips, sweet as honey, had the maiden … the graceful mounds of fair, white breasts, with a covering of golden interlacement]” (A Bhfuil, poem 39). Once their respective fairies “badd [them] loue her deare” (I.IX.14.2), they depart before the morning, leaving the men behind to search for her or await her return. Arthur seeks “her out with labor, and long tyne” (I.IX.15.7), while Ó hUiginn’s speaker awaits his fairy’s return and wishes not to be merely a visitor to her lands:
22. D'éis na mná d'imtheacht uainne gan bheith im' fear éanuaire dá madh éidir é 'na fonn is sé budh éigin orom. [“Since the woman departed from us, I would fain, if it were possible, be not merely a sojourner, in her land”] (A Bhfuil, poem 39)
While “Néall mná síthe sunn aréir” and Arthur’s vision both follow the generic narrative structure of a fairy vision, Ó hUiginn’s poems also engage with other aspects of the sidhe which resonate with the fairylore of Book II. Edwin Greenlaw has argued that Guyon’s trials are full of suggestions of the Celtic faerie.13 Phaedria and Acrasia are fairies who tempt and enchant, while Mammon is the old man who traditionally defends a fairy hill and hoards his wealth.14 Acrasia, Phaedria, and Ó hUiginn’s sídh all enchant using music and flowers. Acrasia’s seductive powers are tied to her sexuality with her “snowy brest [that] was bare to ready spoyle” and her “hungry eies” (II.XII.78.1-2), while Ó hUiginn’s sídh’s enchantments come from her “maidan-like” purity:
4. Béal tana nár mhillse mil ar ghné ógróis 'gon inghin; gach glór róirighin do ráidh fa lór d'fóiridhin easláin. [“Slender lips, sweet as honey, had the maiden, with the hue of a budding rose; every gentle utterance of hers was enough to heal the ailing.”] (A Bhfuil, poem 39)
Despite these differences, all three fairies’ powers are emphasized in their ability to successfully beguile men—Phaedria enchanting Cymochles, Acrasia seducing Verdant, and Ó hUiginn recalling how the sidhe charmed Connla the Red: “Nó an tú do chealg Connla Ruadh/ ó sluagh Breagh na mbeann bhfliuchfuar [Or didst thou beguile Connla the Red, from the host of Banbha of the cold, wet summits]” (A Bhfuil, poem 40). In both Spenser’s and Ó hUiginn’s poems, the protagonist or the speaker resist the temptation of the fairies, again a key element of Celtic fairylore, and an aspect of the tradition that Ó hUiginn celebrates when he recalls the story of Murrough (the son of Brian Boru), who was often brought to fairy mansions but continuously rejected offers of riches, love, and fame: “Nó an tú táinig eacht oile, / go longphort Bhriain Bhóroimhe, / do bhreith Mhurchaidh tar Moir Meann [Or art thou she who came another time to the camp of Brian Bóroimhe, to bear Murrough across the Irish Sea]” (A Bhfuil, poem 40). Greenlaw argues that Guyon’s trials have a similar nature to the story of Murrough.15 If we consider this in conjunction with Clare Carroll’s argument that Spenser “completely reworked Gaelic sources to establish fresh myths,” then it is possible that the similarities between Spenser’s and Ó hUiginn’s poetry occur as Spenser is drawing on Celtic fairylore myths, such as Murrough’s trials, and transmogrifying them for his own means, while Ó hUiginn, embedded in the bardic tradition of citing Celtic mythology, is openly celebrating them.16
This key difference in their representation of the sidhe comes to the fore upon considering the other meaning of the word, mound-dweller, and the way that each poet perceives fairy mounds. For Ó hUiginn they are places of cultural importance, deep history, and beauty such as the “Síth Bhóinne Breagh” [the fairy mounds of the Boyne] or “Síth bhláith tirmleapthaigh Thruim” [warm-couched mound of Trim] (A Bhfuil, poem 40). In The Faerie Queene, however, these spaces are “waste and void” (II.VI.11.3) like Phaedria’s island, or need to be destroyed like Acrasia’s Bower of Bliss. They are places inhabited by degenerates that need to be resisted or reformed.17 In both instances, the fairy mounds could signify Ireland, but while Ó hUiginn celebrates these places, from Spenser’s perspective they need to be cultivated and transformed, a view that becomes explicitly clear by the proem to Book V.
While the application of network algorithms to prosoprographical networks can bring different authors together, the bibliographical networks can bring specific texts into one another’s orbit. In their base form, the bibliographical networks are bi-modal networks in that they connect two distinct types of entities: people and texts. Yet it is possible to project the network to a unimodal state by condensing along the vertex—that is, by connecting texts if they share a person in common. This can be applied to various elements of MACMORRIS’s overarching dataset: either to the texts drawn from the E.S.T.C. in isolation (to see which texts from the English literary tradition are in the same orbit); to the B.P.D. in isolation (to see which bardic poems are in the same orbit as one other); or to a union network of the two (to see which texts are from different linguistic traditions but share a similar role or orbit in the network). When the network algorithms are applied to the E.S.T.C. network, Spenser’s texts map fairly comprehensively onto the prosoprographical Louvain community of which Spenser is a part, and the comparisons remain colonial. His texts share a community with texts by Bryskett, Ralegh, Rich, Churchyard, and Fynes Moryson (1566–1630), to name but a few, and the network profile of The Faerie Queene is similar to that of Rich’s The excellency of good women (1613) and John Davies’ (1569–1626) Nosce teipsum (1599).
Alternately, the union network (which combines the bibliographic data from the E.S.T.C. and the B.P.D.) moves away from the implicit habit of comparing Spenser’s works to those of other colonial writers. When we apply the combination of degree, betweenness and eigenvector centralities to this network and look for texts with a similar network profile to The Faerie Queene, a slightly different picture emerges. Spenser’s magnum opus still has a network profile similar to other works by colonial writers within his orbit. These include Churchyard’s A scourge for rebels (1584) and John Harrington’s (1560–1612) translation of Ludovico Ariosto’s Orlando Furioso (1591).18 Furthermore, the results once more bring Tadhg Dall Ó hUiginn into consideration with three of his poems sharing a similar network profile to The Faerie Queene. These are poems written to Aodh Ó Domhnaill entitled “Dia do bheatha, a mheic Mhaghnais” [Welcome, thou son of Manus], “Maighean díoghla Druim Lighean” [Drumleene is a precinct of vengeance], and “Molfaid Conallaigh clann Táil” [The race of Conall will praise the children of Tál].
Considering these poems are panegyrics eulogising Ó Domhnaill, it is possible to compare Ó hUiginn’s configuration of his patron with how Spenser eulogizes and represents Gloriana and her knights. But a more direct comparison, free from the allegory, is to compare these praise poems to some of Spenser’s dedicatory sonnets, including the sonnet to John Norris (c.1547–97). Both poets praise their addressees for their “warlike prowesse and manly courage,” and ability to triumph over their foes: “'gar roimhionca buadh bhiodhbhadh” [that is most wont to triumph over the foe] (A Bhfuil, poem 4).19 Additionally, Spenser’s listing of Norris’s successes in “sad Belgicke” (DS13.10) and on “Lusitanian soile” (DS13.12) echoes the caithréim, or battle roll often seen in Gaelic praise poetry. However, their respective poems both offer alternate perspectives on the colonization of Munster. Spenser praises Norris for his role in the defeat of the Desmond rebellion as payment for the “prize” of Kilcolman already bestowed on Spenser: “Who euer gaue more honourable prize / To the sweet Muse, then did the Martiall crew.” While Spenser has benefitted from the redistribution of land in Munster, the Gaelic poets have not—because in “Molfaid Conallaigh clann Táil”, Munster is no longer a safe place for poets:
9. Smacht Gall i Mumhain Mheic Con, tarla dhó — dia do phudhor — do bhreith ar ollamh nUladh, ar mbeith ollamh n-ealadhan. 10.File Í Dhomhnaill Dúin na nGall le smacht adhuathmhar eachtrann, gá dtám? — acht torchuinr bu dheas, gur chomhchuir ár na n-éigeas. 11.Ansmacht Gall dóibh fo deara siad d'fulang a oidheadha; [“And it befell—most harmfully—that the foreign rulers of Mac Con's Munster arrested the poet of Ulster when he was perfected in his art. 10. The poet of O’Donnell of Donegal, by the dreadful rule of foreigners— to be brief, he was slain in the south, which caused the simultaneous ruin of the poets. 11. It was foreign tyranny that caused them to suffer his murder;”] (A Bhfuil, poem 4)
While the network profiles bring the poetry of Ó hUiginn into the same sphere as Spenser’s, the application of the Louvain algorithm on the bibliographic network places four other bardic poems in the same community as The Faerie Queene. Three of them are poems to Thomas Butler, 10th Earl of Ormond (1531-1614), for whom Spenser also wrote a dedicatory sonnet. These poems are rich in comparative material and will be examined in forthcoming research by me and Pat Palmer. The final poem is an anonymous eulogy, dated to the late sixteenth century, in the looser amhrán or song-metre to Elizabeth I entitled “I n-ainm an Airdmhic doghni grasa” [In the name of the great son who makes grace]. The poet introduces her as the fifth Tudor prince, “is í dar liomsa an cuigeadh prionnsa,” before going on to praise her beauty, and compare her to Caesar, Hector, and King Arthur. The poet then engages in a caithréim, or battle roll, that emphasizes the might of her navy, giving several examples of recent successes including in Flanders. Because of its dedication to Elizabeth, and its Arthurian and classical allusions, the poem has rich potential for comparative analysis. Indeed, the results of these algorithmic experiments have encouraged one of the members of MACMORRIS to work on the first critical edition and translation of this poem, so that this comparative work can get underway.
Returning to my initial question—can we use network algorithms to enrich or decolonize approaches to reading Spenser—the answer is a “yes, but.” As I have stressed throughout, algorithms can analyze only the data that is inputted. If the material is exclusively anglophone and colonial, then the answer is no. Network algorithms are not generative and will not pluck an answer from the ether; in such cases, the results will only connect Spenser to those with whom he has always been compared. Yet, if we use data that breaks away from an Anglocentric bias and incorporates the prosopography and bibliography of those colonized as well, we get a different answer—a scenario I explore more thoroughly in my forthcoming research on decolonial networks in early modern Ireland. In relation to Spenser, however, the algorithmic results of the Louvain method and the combination of network statistics situates him in the wider polyphony of contexts that make up early modern Ireland. They point us towards other Munster poems, other exemplary poets, and other eulogies to Elizabeth that present alternative contexts, alternative points of view, and alternative literary tropes than those upon which Spenser draws. As has been shown through this comparative reading of Spenser’s and Ó hUiginn’s poetry, reading Spenser alongside this literary record requires us actively to consider how we comparatively read texts from discrete linguistic and cultural traditions in shared geographical and temporal space. This not only requires analyzing similarities and differences in poetics and hermeneutics, but also considering critical historical contexts, power relations, colonial impetuses, and even potential obfuscations of literary traditions. Above all, it reminds us to move beyond the Anglophone canon and to remember that Spenser is just one perspective in an infinitely complex polyvocal landscape.