Joint and multi-authored publication patterns in the Digital Humanities
My colleague Oliver Duke-Williams and I had a paper on ‘Joint and multi-authored publication patterns in the Digital Humanities’ accepted to the recent Digital Humanities conference in Nebraska. After the talk a number of people contacted us to say that our findings would be useful for them when they sit on tenure, promotion and hiring committees because they could help to contextualise the publishing patterns of Digital Humanists within the wider Academy. So, we are making our initial findings available here. Please note that we are working on a paper about this topic that we will submit for peer review in the coming months so these findings are very much preliminary and will be extended further and possibly revised (slightly). Note that we also have a number of findings about the gender of authors, especially in terms of collaboration patterns. These will be included in the full paper and we say a brief word on genders below.
1.0 Research Context
A frequently claimed hallmark of Digital Humanities is its emphasis on collaborative work and joint publication (see, for example, Moulin, Nyhan et al 2012; Koh 2012; and Deegan and McCarty 2012). Is there any mismatch between the way that the field describes itself and what we find when we examine the evidence of publication patterns and practices? Furthermore, have publication patterns changed since the first journal of the field, Computers and the Humanities, was established in 1966? Has joint publication become more or less common or have the proportions of jointly published articles remained the same? Also, how do such patterns compare with other disciplines of the Humanities?
2.0 Methodology
Using Zotero we extracted bibliographical metadata from Computers and the Humanities (Chum) (1966 – 2004); Literary and Linguistic Computing (LLC) (1986-2011); and Digital Humanities Quarterly (DHQ) (2007-2011) and then exported this to Excel for initial viewing. As far as possible the data was cleaned and regularised e.g. a canonical form of personal names chosen where slight differences existed such as E.G. Wills and Edward G. Wills. The cleaned data were then imported into an SQL database and sorted into groups based on the number of authors. The annual observed frequencies of papers with n authors were then calculated. For each group, a linear regression was calculated in order to determine within a given journal whether the incidence of n-authored papers had changed over time.
For each journal the data were also processed so that dual-authored papers could be analysed using a connectivity index (Bell et al 2002) to determine the extent to which the pool of authors contributing to a given journal were interconnected. A connectivity index was constructed both on a journal-wide basis and on a per-author basis, allowing the distribution of ‘well-connected’ authors to be compared within and between journals.
A look-up table was used in the database to map different variant spellings of a person’s name to a single canonical version. This occurs quite commonly, and is more common for people who have published heavily over a long time period: in the worst cases, there were up to five different spellings of a person’s name.
This table had to be prepared by hand, and this obviously risks both false pairings (two very similar names which are not, in fact, the same person) and omissions (two names which should map to the same person but are treated as separate people). As we check our results, we’ll also check this table closely for both sorts of errors. Neither sort of error should affect the analysis of the frequency of papers with different numbers of authors, but might affect some of the other analyses
3.0 Computers and the Humanities (1966-2004): findings
In CHum we see that the trend over time is for single authorship to predominate and hold steady; there is a significant increase in dual and triple authored papers
Here are the correlations: Note that
**. Correlation is significant at the 0.01 level (2-tailed)
*. Correlation is significant at the 0.05 level (2-tailed)
3.1 Literary and Linguistic Computing: findings
Here are the figures for the correlations. Note that
**. Correlation is significant at the 0.01 level (2-tailed)
*. Correlation is significant at the 0.05 level (2-tailed)
3.2 Annals of the Association of American Geographers (1966-2013): findings
In order to build a comparative context for the Digital Humanities figures we carried out a literature review of trends in the broader area of the Humanities. We couldn’t find many studies that looked quantitatively at collaborative publishing patterns among Humanities scholars. So, we carried out an analysis of an exemplary Humanities journal: the Annals of the Association of American Geographers during the years from 1966-2013. In the full paper we will include another Humanities journal in order to build a more robust comparative context. Here are our initial findings:
• Regarding the excellent point made by Isabel Galina
Stats shown by @juliannenyhan on publication patterns. DH maybe not as open and collaborative as self-perceived? #DH2013
— Isabel Galina (@igalina) July 19, 2013
We can only say #watchthisspace
• Finally, Oliver has made available some data tables which summarise some of the data with which we have been working.
The tables show author, number of papers and publication years for each journal. The publication years are given an asterisk where a paper was co-authored by two or more people. Clicking on the header for a column will cause the data to be sorted and redisplayed on the basis of the column contents. You can find them here.
Bulger, Monica, et al. Reinventing research? Information practices in the humanities. Research Information Network, UK, April 2011.
Deegan, Marilyn, and Willard McCarty. 2012. Collaborative research in the digital humanities : a volume in honour of Harold Short, on the occasion of his 65th birthday and his retirement, September 2010. Farnham: Ashgate.
Engels, T.C.E., Ossenblook, T.L.B. & Spruyt, E.H.J. (2012). Changing publication patterns in the Social Sciences and Humanities, 2000-2009. Scientometrics
Koh, Adeline. 2012. ‘The Challenges of Digital Scholarship’. ProfHacker, January 25. http://chronicle.com/blogs/profhacker/the-challenges-of-digital-scholarship/38103
Moulin, Claudine and Julianne Nyhan et al. (2011) Research Infrastructures in the Digital Humanities ESF: Science Policy Briefing
I’d be intrigued to see a relative weight of the importance of collaboratively authored works. The most important paper in LLC is probably Delta (single author), but the Weingart and Jorgensen paper (dual author) often crops up in various contexts. Also worth pointing out – back in the day Opas and Tweedie were a dynamo, pushing out papers on pretty much everything from popular culture to dialect evolution. People like Hockey and Burrows do exist, but they are a relative rarity. Also worth pointing out – computer science folks usually work in teams, so this may be a marker for disciplinary boundaries. A final point – collaboration is now more finely-grained. The stylo package means that people can use Eder’s code with only a citation: co-author credit is not necessary.