Author: cottonvespasian

Who’s Afraid of all but Beowulf?

~ cottonvespasian ~ Leave a comment

I wanted to make the claim on Monday that large parts of the 3m-word surviving Old English corpus are neglected. Large parts of the 3m-word surviving Old English corpus are neglected, but – being someone who likes to quantify things – I began wondering if there was a way to calculate how much Old English criticism has focused on the most canonical texts, like Beowulf. In the back of my mind was Andrew Goldstone and Ted Underwood’s mind-blowing distant reading of seven major literary history journals between 1887 and 2013 in ‘The Quiet Transformations of Literary Studies’, with its extraordinary visualisation of the topics that have preoccupied literary scholars in the last century:

While I haven’t produced anything on a par with this, I have found a way to calculate how much Old English criticism focuses on the most canonical texts, and – incidental to that – a way to look at the disciplinary health of Old English studies relative to academia more broadly. It has also given birth to this blog – a new venue to publish things that are interesting (and I’d like to be able to cite) but which I don’t have the time to work up as fully-fledged articles.

My first idea for a dataset was the Old English Newsletter Bibliography, heroically kept online by Stephen Harris and shared by him with my PhD student, Claire, a couple of years back. But Claire had a job application to write, and the OEN Bibliography only extends to 2010, while I wanted to make a claim about present-day Old English studies. So I started playing with Brepols’ (pay-walled) International Medieval Bibliography (originally founded in 1967), and looking at how a random selection of articles on Old English had been manually indexed there. If all articles on Old English were indexed to a single overarching category, life would have been easy. Instead, articles are primarily indexed by Discipline (e. g. ‘Literature’) and Area (e. g. ‘British Isles’), and – as we all know – not all literature from the British Isles in the early Middle Ages was in English. Consequently, I decided I would need to utilise the index terms for both Subject and Persons, Families, Texts, for Subject included ‘Old English’ and Persons, Families and Texts often included text-based index terms including ‘OE’ (e. g. in ‘Maxims, OE poem’). ‘OE’ yields 4,255 hits in IMB, and ‘Old English’ 3,466. While the IMB interface has a button that says ‘Export All’, the resulting download is in fact capped at 500 entries, so I had to extract the data in batches. Because some articles were indexed with both search terms, I deduplicated on the basis of the IMB identifiers given to each item (e. g. IMB (a1137767)). That left a dataset of 6,774 publications.

This list is certainly not a complete inventory of publications on Old English between 1967 and the present day. First and foremost, the IMB until recently almost totally excluded monographs and primarily indexes articles in journals or edited collections. Presumably, too, because of the inherent difficulty of comprehensiveness, they missed some articles on Old English in obscure publications. More seriously, some publications on Old English are not indexed as either ‘OE’ or ‘Old English’. This seems to be particularly the case with prose works. For instance, of 690 publications indexed ‘Ælfric, abbot of Eynsham’, 162 (23.5%) do not contain ‘OE’ or ‘Old English’ as index terms. I should probably have searched the IMB for all known Old English author and work names, but that would be quite exhausting (and still not necessarily ensure exhaustiveness). Despite covering a narrower period (1970-2010), the OEN Bibliography is certainly more extensive than my dataset from IMB, with 23,700+ rows in the csv file Stephen shared with Claire, but it includes monographs and other publication types, and it looks like it would take a lot of work to make any kind of confident statement about its relative coverage. So I rolled with what I had.

One bit of minor data wrangling was necessary before producing any visualisations. In the IMB export, not all publications have a date in the date column; however, they are listed in (reverse) chronological order, so any publication that appears in a row between one from year XXXX and another from the same year must also be from year XXXX. Where the row was between one from year XXXX and one from YYYY, I manually verified the date. For computation purposes, I treated publications from XXXX (for YYYY) as being published in year XXXX (my thinking being that even in the case of an article published in ZZZZ, the author may in fact have submitted it five or ten years beforehand; the editors were taking some zzzzs, presumably) and publications in journal volumes for XXXX-YYYY as having been published in the later year. I also excluded any publications that were not journal articles or book chapters, given the under-representation of other publication types in IMB. This gave me raw counts of publications on Old English per year. Inspecting these results suggested IMB coverage at the extremes – 1966 and 1967 at the beginning and  2025 and 2026 at the end – is a bit patchy, so I discounted these years. I also applied some rudimentary smoothing, by averaging each year’s production across the two years before and after it. That yielded this graph:

So there are about twice as many articles and book chapters being written about Old English now than there were in the 1970s: horray! Quite what explains the peaks in productivity in the mid 1990s and late 2000s, I’m not sure. We’d need to bear in mind the possibility that the peaks are not peaks in publication, but peaks in the assiduousness of the IMB’s compilers, and absent Brepols giving me the whole IMB to play with, I can’t think of a way to test that possibility.

But is the health of the discipline actually as rosy as this graph suggests? More and more journals are being established and thus, presumably, more and more articles being published (see Brian McGill’s post, ‘The State of Academic Publishing in 3 Graphs, 6 Trends, and 4 Thoughts’). There may be better datasets about total academic publications over time out there, but a brief search led me to To & Yu’s ‘Rise in Higher Education Researchers and Academic Publications’ from 2020, which utilised – inter alia – figures for total numbers of publications indexed by Scopus for the period 1980-2018, and conveniently made these available for download among the article’s supplementary materials. Applying the same rudimentary smoothing to this data yields parallel data to IMB for 1982-2016. Taking 1982 as 1 and calculating the number of articles published each year relative to that enables the datasets to be compared over time:

So the picture is actually not quite so rosy. The number of publications on OE has been rising over time, but it has not kept pace with overall publication trends. While in rude health to 1995, publication on Old English has remained essentially stable since then, while academic publication in general increased almost threefold in the 20 years that followed.

But what of the topics those publications on Old English have addressed? Have they changed? The ‘Export all’ function in IMB does not yield the topics under which each publication was indexed, but it does give the titles. I therefore decided to concordance these, having first deleted all non-capitalised words (as a kind of rudimentary alternative to using a stop list to discount the articles, prepositions etc that reveal nothing about a publication’s topic). That yielded a list of 4,245 words occuring between them 23,496 times in the titles of the 6,774 publications in my IMB-derived dataset. Two thirds of these are hapaxes (and have some possibility of serving as the basis for a round at the next quiz at the IMC: can you name the only article on Old English which refers to Catullus in its title?). The top 1% of title words by ranked frequency are also the only ones that occur over 20 times per 10,000 title words, and these 42 include the names of seventeen canonical Old English works and authors: Beowulf (1376 instances), [Battle of] Maldon (130), Genesis [A/B] (129),[Dream of the] Rood (124), Christ [I / II / III / & Satan] (122), Exodus (120), Ælfric (116), Andreas (99), Wanderer (83), Riddle [XX] (73), [OE] Bede (68), Guthlac [A/B] (64), Seafarer (61), Wife[’s Lament] (59), Judith (56), [King] Alfred (55) and Wulf [and Eadwacer] (51).

To see how prevalent each of these canonical works and authors has been in Old English scholarship over time, I identified the corresponding IMB Persons, Families and Texts labels for each. In some cases this was just one label, e. g. ‘Battle of Maldon, OE Poem’ but in others it was more than one, e. g. for Genesis ‘Genesis, OE Poem’, ‘Genesis, OE poem – Genesis A’ and ‘Genesis, OE Poem – Genesis B’. This yielded a list of 4,265 publications using these index terms. Having ensured each was associated by a single year, and – as with the broader OE dataset above – counting only journal articles and essays in collected volumes, and conducting the same rudimentary smoothing, we can look at what portion of the scholarship on Old English over the last fifty years has focused on Beowulf:

An important caveat before we interpret this graph is that articles about Beowulf are not necessarily articles exclusively about Beowulf, but rather articles which treat Beowulf to a degree significant enough for the IMB’s bibliographers to notice; they may also treat other texts. With that in mind, we can say that interest in Beowulf peaked in the early 1980s, when almost 50% of articles on Old English had something substantive to say about it. Since the late 1980s, about a quarter of published articles have been concerned with Beowulf.

But what about canonical texts tout court? I grouped the seventeen canonical texts and authors on which I collected data into prose and verse, yielding three prose (Ælfric, Alfred and the OE Bede) and fourteen verse texts. Because some articles concern multiple canonical texts, e. g. Bourquin’s ‘Lexis and Deixis of the Hero in Old English Poetry’, which discusses Andreas, Guthlac, Judith, Beowulf, Battle of Maldon, Dream of the Rood, Exodus and Christ poems – Christ indeed! – I deduplicated so each article in the dataset focused on a canonical verse text counted only once, and each article on a canonical prose text counted only once. Furthermore, because a few articles (51 to be precise) dealt with a canonical prose text and a canonical verse text, when calculating the total number of articles published per year on canonical articles, I counted these articles once only. This means that the total number of articles published on canonical texts is very occasionally less than the sum of the number of articles published on prose and on verse canonical texts (e. g. in 1976, I have 58 articles total on canonical texts, but 6 on prose and 53 on verse, because Paul Szarmach’s summary of the Old English Division meeting at the MLA for OEN treated papers that concerned both canonical prose and verse texts). Bearing that in mind, we come to my final graph:

This graph suggests that the proportion of research on Old English concerned with canonical texts was, as of 2022, about 50%. That figure has remained more or less stable since the late 1980s, having peaked at nearly 90% in 1980. Scholars have been much more interested in verse than prose over the last fifty years, with canonical verse texts commanding about three times as much attention as prose texts in 2022. Interest in canonical prose texts almost tripled between the late 1970s and the early 2000s, but since then has almost halved (though there inchoate signs of an uptick since 2020).

So, to answer my own question, the IMB dataset suggests that about half of the criticism on Old English at the moment is on the seventeen most canonical texts. Or, to put it another way, over the last fifty years, 2,765 articles out of 6,352 on Old English have focused on fourteen verse texts; that is, 44% of publications on Old English have focused on 2% of the corpus by wordcount. If that’s not an incitement to go and look at one of the other 3,052 texts in DOEC, I don’t know what is.

***

There’s probably a lot more that could be done with this dataset, for instance to explore the no less than nineteen languages in which it indicates scholars have published on Old English. It is, I hope, needless to say that nothing above is intended as a criticism of anyone who works on canonical Old English texts. I well understand the pressure that working in an English department might create to focus on things that colleagues who specialise in other period might understand, like poetry (indeed, one of the few bits of advice one of my PhD supervisors gave me, when my first article (on an obscure, and probably non-existent, East Anglian saint) came out, was to try to publish something on something canonical soon so I’d be comprehensible on the job market). The aim of the article is rather to hold a mirror up to what we publish on, without opining (too much) about whether that is right or wrong. If anyone would be interested in collaborating to write this up properly, let me know and I’m happy to share the provisional data.

Corpus philology: Using the Dictionary of Old English to get bigger data for Old English spelling variation

~ cottonvespasian

This article presents a methodology for obtaining large datasets for the spelling of individual phonological segments in Old English texts, based on searching the Dictionary of Old English Corpus for the attested spellings listed in the Dictionary of Old English A-H. It exemplifies this ‘corpus philology’ through a study of 216,526 spellings for words beginning with h followed by a vowel, using a variety of techniques to evaluate the methodology’s precision and recall, which are calculated as very high for <h->initial spellings (precision 100% precision, recall 92.1%) and moderate, but still usable, for <h->less spellings (precision 85.5%, recall 58.3%). Data for fourteen other segments related to the behaviour of h- in Old English is presented in the Supplementary Materials that complement the paper online. This dataset of 379,484 spellings from 2,605 Old English texts is shown to seriously problematize the findings of traditional philology, the conclusions of which are in contrast based on only a handful of spellings from a few texts, and to have the potential to radically enhance our understanding of the literary and linguistic histories of English.

Open Access available here.

Reference: Mark Faulkner, ‘Corpus philology: Using the Dictionary of Old English to get bigger data for Old English spelling variation’, Digital Scholarship in the Humanities. Advance access published 11 October 2023.

Corpus Philology, Big Dating and Bottom-Up Periodisation

~ cottonvespasian

Argues, with Crane, Bamman and Jones 2013, that philology can be seen as a special case of corpus linguistics, showcasing a range of techniques that allow for the contextualisation of linguistic variants in Old English texts using corpus-linguistic techniques. As a case study, it attempt to date two English versions of Sawyer 136 (King Offa for St Albans, ostensibly from 793), suggesting an eleventh-century date for its forgery is feasible. The conclusion delineates some broader possibilities that derive from these corpus-philological methods, including using cluster analysis on linguistic profiles of texts and applying techniques like Variability-Based Neighbour Clustering to ascertain where to place the boundary between Old and Middle English.

Pre-print available here.

Reference: Mark Faulkner, ‘Corpus Philology, Big Dating and Bottom-Up Periodisation’, in Dark Archives I: Voyages into the Medieval Unread and Unreadable 2019-2021 eds. Stephen A. Pink and Anthony Lappin, Medium Ævum Monographs n. s. 43 (Oxford, 2023), 285-313.

Using Manuscript Books as a Source for Medieval Culture

~ cottonvespasian

A case study of what can be gleaned about how medieval texts were written and read from the close study of a single manuscript, in this case a copy of Lydgate’s Fall of Princes in the Newberry Library Chicago, which is available in digitised form alongside the case study.

Reference: Mark Faulkner, ‘Using Manuscript Books as a Source for Medieval Culture’, in Research Methods Primary Sources (Marlborough: Adam Matthew Digital, 2021).

The Old English Bede: a new source for the F-version of the Anglo-Saxon Chronicle

~ cottonvespasian

Identifies the Old English Bede as the main source for annal 995 of the post-Conquest Domitian Bilingual, where its compiler puts into the mouths of some fictitious wise men a spurious account of Christ Church’s history reaching back to the time of Augustine, which they claim is ‘swa þu ræddan miht on Ystoria Anglorum’.

Preprint available here.

Reference: Mark Faulkner, ‘The Old English Bede: a new source for the F-version of the Anglo-Saxon Chronicle‘, Medium Aevum 90.2 (2021), 217-229.

A New Literary History of the Long Twelfth Century: Language and Literature between Old and Middle English

~ cottonvespasian

When William the Conqueror was crowned in 1066, to the Norman guards at the coronation English was an ‘unknown tongue’. My New Literary History of the Long Twelfth Century focuses on post-Conquest English as the unknown tongue of English literary and linguistic history, where it habitually falls between the subperiods of ‘Old’ and ‘Middle’ English. The book’s approach spans the disciplines of literature and linguistics. Both have always seen the long twelfth century as pivotal but essentially inscrutable. Linguistically, the transition from Old English to Middle English in the course of the period has been described as ‘the most dramatic change in the English language’, but this change remains seriously underexamined. In literary studies, the connection between the body of texts aggregated as ‘Old English’ literature and those labelled as ‘Middle English’ remains tantalising. Texts composed in the long twelfth century, produced at the intersection of these two periods, invite us to consider scholarship’s construction of ‘Old’ English, ‘Middle’ English and thereby the entire body of medieval texts. The book offers a literary history of English language texts in the long twelfth century, considering texts first and foremost as linguistic objects and the various approaches it adopts to understanding their language are described in a methodological chapter that follows the introduction. Part I of the book establishes the affordances of English in the long twelfth century. Part II examines how English was used during this period in three different genres: documents, histories and sermons.

Reference: Mark Faulkner, A New Literary History of the Long Twelfth Century: Language and Literature between Old and Middle English (Cambridge: Cambridge University Press, 2022)

Habemus Corpora: Reapproaching Philological Problems in the Age of ‘Big’ Data

~ cottonvespasian

This paper demonstrates the potential of new methodologies for using existing corpora of medieval English to better contextualise linguistic variants, a major task of philology, and a key underpinning of our ability to answer major literary-historical questions, like when, where and to what purpose medieval texts and manuscripts were produced. The primary focus of the article is the assistance these methods can offer in dating the composition of texts, which it illustrates with a case study of the “Old” English Life of St Neot, uniquely preserved in the mid-twelfth-century South-Eastern homiliary, London, British Library, Cotton Vespasian D.xiv, fols. 4–169. While the Life has recently been dated around 1100, examining its orthography, lexis, syntax and style alongside that of all other English-language texts surviving from before 1150 using new techniques for searching the Dictionary of Old English Corpus suggests it is very unlikely to be this late. The article closes with some reflections on what book-historical research should prioritise as it further evolves into the digital age.

Published article available now.

Reference: Mark Faulkner, ‘Habemus Corpora: Reapproaching Philological Problems in the Age of ‘Big’ Data’, Anglia: Zeitschrift für Englische Philologie 139 (2021), 94-127.

The final publication is available at www.degruyter.com.

Medieval Manuscripts

~ cottonvespasian

An annotated bibliography of 100+ items for anyone interested in working more extensively with medieval manuscripts. As dictated by its publication in the section of Oxford Bibliographies Online devoted to British and Irish Literature, the focus is on Latin, English and French manuscripts from the insular world, c. 500-1500 (manuscripts in Celtic languages I largely left to the experts!).

Reference: Mark Faulkner, ‘Medieval Manuscripts’, Oxford Bibliographies Online (2019).

English

~ cottonvespasian

Co-written with my former Sheffield colleague, Joan Beal, this is an introduction to the role of language contact in the shaping of the English language, which particularly considers contact with Celtic, Norse and French from the beginnings to the present day.

Reference: Joan Beal and Mark Faulkner, ‘English’, in The Oxford Handbook of Language Contact ed. Anthony Grant (Oxford: Oxford University Press, 2020), pp. 374-387.

Quantifying the Consistency of ‘Standard’ Old English Spelling

~ cottonvespasian

This paper presents two large datasets, one of over 19,000 morphemes from 100 texts, the other of over 91,000 spellings from 941 texts across 198 manuscripts for one phonological segment, with a view to providing an empirical basis for discussions of the homogeneity with which the vernacular was written in late Anglo‐Saxon England. These show the infinitive morpheme was spelt <‐an> 96.1 per cent of the time and the diphthong /æa, æ:a/ was written <ea> 96.5 per cent of the time. Such consistency, over time and across dialect boundaries, suggests that recent scepticism about the existence of a homogenous, conservative, supraregional variety of written Old English, perhaps a ‘standard’ Old English, is unwarranted.

Download Pre-print

Reference: Mark Faulkner, ‘Quantifying the Consistency of ‘Standard’ Old English’, Transactions of the Philological Society 118 (2020), 192-205.