Throughout this academic year, fourteen colleagues and I are engaging in a faculty/staff learning community investigating open scholarship. This fall semester we have built our shared understanding and our co-constructed meaning around this concept of open. During the spring semester ahead, we will pivot toward more practical and applied work in the form of individual projects.
As this semester speeds toward its end, I admit to a difficult time finding my footing. Our collective undertaking is admittedly broad, but even so I believe my difficulty lies with the stubborn ambiguity of ‘open’ and my slow recognition that ‘open’ is a problematic and unfortunate framework for discussing accessible scholarship and participatory publics. So I offer this post as a way for me to catch up to other participants already doing their great work, and to reconsider my understanding within the time that remains.
I have experienced, to greater or lesser degrees, six broad areas of professional and scholarly engagement around some notion of open: open standards (which I’ll sideline here, as they are perhaps the least ambiguous), open source software, open access publishing, open data (and here primarily open geospatial data and statistical data), open educational resources (OER) and what are called open pedagogies and/or open educational practices.
Within any of these broad areas, negotiation of terms and concepts is necessary. For example, understanding the nature of openness of open source software necessitates qualification and classifying distinctions, with both ‘open’ and ‘free’ concepts accommodated within a F/LOSS detente. The same holds for any comprehensive assessment of open access (OA) resources, resulting in the cumbersome and not especially helpful classifications of Green OA, Gold OA, and Black OA to signify which party might be responsible for securing or relinquishing access rights, or in the case of Black OA, a principled ignoring of rights. Consideration of Open Access materials is further complicated by a differentiation between ‘gratis’ access — signifying immediate, persistent, and free access to OA resources, versus ‘libre’ OA — signifying immediate, persistent, and free access along with rights to reuse, modify, and redistribute OA resources.
With respect to Open GIS, complexities converge. An Open GIS community of scholars and researchers need negotiate both open source software ambiguities, as well as those akin to open access resources whenever access and reuse rights apply to geospatial data use, redistribution, or resulting publication.
Within the disciplinary sphere of Education, ‘open’ perpetuates all of these negotiations and introduces new ones. Open Educational Resources (OER) may narrowly mean an open (immediate, persistent, and free) alternative to proprietary texts. But OER may also mean materials designed upon and for use within hypertextual environments. OER may emphasize a community of practice and shared authorship and oversight. One expectation of OER may be a living work, as opposed to a work updated through the release of printed editions, each supplanting its predecessor. OER may be oriented toward the learner (e.g., textbooks), or the practitioner (lesson plans, assignments, assessments). I find openness as applied to OER is especially ambiguous.
The nature of Open Educational Practices and their relationship to and distinction from Open Pedagogies is a matter I hope to address in a future post. In brief, I feel it is safe to assert that considerable ambiguities arise around a notion of ‘open’ classrooms, learning environments, and assignments. The positioning of the instructor in relation to the learner, the positioning of one learner in relation to another, and the position of either with respect to time and/or distance are all greatly complicated by imagining the practices of teaching and learning while ‘open’.
This is all to establish that when I read or hear ‘open’ I’m often not sure what to do with the word. Even placed adjacent to scholarship, I struggle for a toehold. I’ve felt that my business within this learning community could not proceed until I had a more fixed understanding of what I mean when I use ‘open’ to represent one or more inter-related concepts.
Open as Keyword
For help, I turn to Raymond Williams and specifically to his book, Keywords: A Vocabulary of Culture and Society. Williams did not write directly upon the subject of ‘open’. Instead, I borrow from his practice of critical etymology (perhaps cultural philology?) and look very carefully at ‘open’ as a way to organize and understand the many and increasing ways in which it is used. I also scan many other derivative works to see if anyone has given ‘open’ keyword treatment.
When Williams returned to Cambridge University after his studies were interrupted by World War II, he and other students sharing that experience found it difficult to engage around particular words and concepts. Williams states he felt these two groups of students, those who participated in the war as soldiers and those who did not, seemed to almost speak different languages. Certain words, as symbols of important or contested concepts, seemed to glow brighter than others. It was this recognition that prompted Williams to begin assembling a vocabulary of terms he felt had shifted, in part due to the great cultural and social upheaval of a world at war. That work would culminate in the publication in 1958 of Culture and Society 1780-1950, a cornerstone of the emergent field of Cultural Studies. In his introduction to that work, Williams identifies five words, in particular, requiring careful attention. These are industry, democracy, class, art, and culture. Keywords initially was intended to be both preparation for and an appendix to Culture and Society. It was published as its own expanded work in 1976.
I am quick to state that I don’t possess the training or capacity to apply a rigorous philological analysis of the word ‘open’ as Williams might. But I did find referencing the OED to be tremendously helpful to me. The first thing I observed about the word is its entry requires ten 3-columned pages! Open is a workhorse of a word, and one borrowed from Old German as oppen in early Old English . ‘Open’ or some progenitor has been with us for a very long time. The Proto-Germanic *uppa, and a Proto-Indo European root,*upo, both signify taking something from below, where it is concealed, to above, where it might be seen. It is from this directional primitive — moving upward from where an object is underneath and concealed, past a barrier, onto a plane where an object can be observed — that all other senses of ‘open’ derive.
There are dozens of OED entries extending back to early Old English representing verb, noun, and adjective parts of speech. The usefulness of ‘open’ is tough to match in English. Some prime senses of ‘open’ include:
a cavity or pocket within earth or rock (cave opening); navigating through obstruction to emerge in cleared space (opening in the woods); acts of parting, fanning, or expanding something hinged (open the gate, the fan); a state (of mind) that is receptive or clear (open to suggestion); to establish a new thing or to permit entrance to it, (school is open, opening night); use of the senses or body faculties, (open your mouth, opened arms, keep your eyes open); manipulating an object (physical or digital) to reveal what is concealed (open an envelope, open the software program). Obviously, there are many others, but these senses of ‘open’ allow sufficient specific understandings of its uses, and help to emphasize the directionality of ‘open’.
In this way, ‘open’ is almost as central to English as its many prepositions. But unlike prepositions that have relinquished a degree of semantic absorbency in favor of their syntactic operation, ‘open’ remains available to proliferating associations and applications. In fact, what makes ‘open’ problematic is not a sense of semantic re-assignment, as Williams observed about ‘culture’ and ‘society’. Rather, ‘open’ in our current context seems almost too yielding and accommodating.
Williams ably traces the evolution of his selected keywords through their appearances in writing, while considering cultural and social changes of the times during which the ‘great’ works were produced. Across everything, Williams applies a Marxist lens, evaluating the production of cultural and social artifacts (within and beyond the English language) as expressions of markets or under influences of class structures. I have not fully reconciled this reliance upon the OED’s high cultural canon with Williams’s New Left sensibility. For that reason, I’ve often wondered what Williams might have made of Information Age tools and methods possible with the ascendence of computation. An analysis of language, and the possibility of informing the creation of keywords using a vast corpus, machine learning, and computational linguistics might have permitted Williams other methods for evaluating relationships between language and culture through time. Specifically, I wonder what Williams might make of tools like WordNet, and FrameNet, or what he might think of the visual representations made possible by tools like the Google Books NGram Viewer.
Google Books NGram Viewer
The Google Books NGram Viewer counts the occurrences of individual words, and plots them over time based upon the publication date of the works in which the appear. It is basically an enormous bag of words with some light treatment, like part of speech tagging, given to each word. The Viewer interface allows for the plotting of the frequency occurrence of a particular word, or the comparison of multiple words over time. It even allows for queries indicating when one word occurs within n-positions of another (‘open’ within one place of ‘source’, for instance). Below, I’ve embedded a couple explorations of ‘open’ in the NGram Viewer as manipulatable frames. Please play around and see what you learn.
The datasets underlying the NGram Viewer are free (available for download, reuse, and manipulation), and covered by a Creative Commons Attribution 3.0 license. While there are concerns about using the Google Books NGram data sets for cultural or lexical analysis due to the nature of the corpus (an overabundance of scientific texts, among other factors), I still wonder what Williams might think of our present ability to visualize term frequency distribution. Here, I love to imagine how these tools might have proven useful to Williams in his analysis of culture and society through his careful consideration of language through time.
WordNet (https://wordnet.princeton.edu/) is a lexical database engineered to permit both human and computer uses. WordNet is open (immediate and persistent) and freely available (does not cost money and permits repurposing via a BSD software license). A search of Github for ‘WordNet’ reveals 897 repositories, each a distinct project or connector incorporating the WordNet database into a natural language processing undertaking. Machines may use WordNet as a traversable network of word meanings for natural language processing. WordNet is particularly helpful for word sense disambiguation tasks such as determining which sense of ‘open’ might be understood by a machine within text or spoken utterances. WordNet organizes not only individual words, but treats word pairs like ‘open up’, for instance, as single entries within its lexical database. These combinations of words are given linkages to other related senses of a word or words, thereby creating a web of semantic relationships that can be traversed by decision-making algorithms. In the same way that second language acquisition reveals eccentricities of meanings, traversing the entry for ‘open’ in WordNet reveals relationships with the ‘heart-to-heart’ (adj), the ‘surface’ (v), and the ‘broadcast-area’ (n) synsets. Where the OED extracts word senses from a canon of classical and contemporary texts, WordNet permits engagement at the corpus level. I wonder what Williams might make of a chance to see how various keywords are used in real time, across, for instance, millions of tweets a second.
FrameNet (https://framenet.icsi.berkeley.edu/fndrupal/) is a project that generates a lexical database useable to both humans and machines. By creating hierarchical semantic frames around actual texts, FrameNet aims to aid machine learning and computational text analysis. Tasks like word-sense disambiguation, for instance, are aided by this network of semantic frames. For instance, when a machine encounters the term ‘key’, evidence of the ‘sound’ frame (the proximity of words like music, listen, pitch, tone, etc.) can help disambiguate from the ‘access’ frame (with instead its adjacent words like lock, closed, door, etc.).
FrameNet is particularly difficult to link to, but I’ve inserted a screen capture of the ‘open.a‘ lexical term contributing to the openness frame.
For the purpose of this post, it is less important to understand how FrameNet works. Instead, the nature of the project (collaborative annotation by humans for the benefit of human and machine understanding), and the open and free availability of its data, are most relevant. Additionally, it becomes evident after exploring the platform that ‘open’, as we see in WordNet and the OED, requires a lot of attention due to its flexibility and utility. What is also important, I believe, is the semantic grounding of ‘open’ or ‘openness’ in the higher-order concept of access.
Williams, Keywords, and the growth of a literary genre
This tour of tools is useful in illustrating various ways in which ‘open’ is applied to the field of Computational Linguistics, and is likely consistent with most 21st Century scholarly contexts in its leveraging of communities of interest and the Internet. However, it may be similar or quite different from other scholarly cultures’ principles of attribution and permissible re-use. In the same way Williams was interested in a small cluster of words and ideas (society, industry, culture, democracy, class, and art), I believe a better understanding of ‘open’ demands similar careful critique and analysis of related words and concepts. Particularly, I believe open, cannot be sufficiently understood without coordinate investigation of access(ible), free(dom), and commons/community, and public. Second-order consideration should be given to the terms software, attribution, and license.
While much of this work seems necessary for my clarity, there is one area where I especially hope this kind of careful consideration may prove illuminating. I am interested in the extent to which access around ‘open’ initiatives is enforced within various economies. Like Williams, I believe grounding analysis in response to mostly market-based factors will be the best and most fruitful course. I believe ‘open’ undertakings are, at their core, a response to markets and economic class constraints. Further, I imagine the various communities and publics forming around and across open initiatives serve to regulate access to the commodities of open labor. Lastly, I believe an examination of ‘prestige’ will help advance understanding around ‘open’ production and its associated gatekeeping. This can take the form of ‘glamour publications’ and the various economies that emerge around scholarly publication. This can also emerge through the performance of production within Open Source Software communities, where some work is privileged, such as writing code, over other labor, such as writing documentation or maintaining community bonds. One or more posts related to this are forthcoming.
Meanwhile, I thought it would be helpful to scan for evidence of an ‘open’ keyword suitable for borrowing or expanding. Since its initial publication, Keywords: A Vocabulary of Culture and Society has spawned an entire genre of Cultural Studies works. Today, titles address keywords for fields as seemingly disparate as Police Administration, Romanticism, Software Studies, and Sound. Most works preserve some degree of Williams’s linkages between related entries, but few preserve as thorough a reliance upon any one work as did Williams on the OED.
Of the many derivatives, a popular editorial choice is to invite individual submissions from authors possessing a particular expertise (see: Keywords for Radicals: The Contested Vocabulary of Late-Capitalist Struggle; Digital Keywords: A Vocabulary of Information Society and Culture; Critical Terms for Media Studies; New Keywords: A Revised Vocabulary of Culture and Society). These titles from NYU Press, are even published as a series.
Some, such as John Pat Leary’s Keywords for the Age of Austerity and the University of Pittsburgh’s Keywords Project are born digital, affording search and hypertextuality not possible in 1976. But Williams, in his own right, presupposes an interrelationship among his selections through cross-referencing and ‘see also’ classifications within entries.
While Digital Keywords contains a placeholder entry for ‘Open’ within its Appendix, only one initiative (also born and maintained online), provides a completed entry for ‘Open’. The MLA Commons project entitled, Digital Pedagogy in the Humanities: Concepts, Models, and Experiments contains 60 keywords generated and written via a community authorship approach. It is here, finally, where specific attention to Open as keyword can be found, thanks to Michael Roy.
In this entry, ‘open’ is a means toward democratic access to knowledge. It requires the open nature of the standards and technologies that comprise the Internet. Ultimately, ‘open’ exists as a form of stasis between oppositional forces such as open versus private, fixed versus rewriteable, and authoritative versus crowdsourced, among others. While orienting ‘open’ towards its implications for pedagogy, I find these three forms of tension to be helpful more broadly.
In Software Studies: A Lexicon, an ‘open source’ entry describes a distinction between source code that is human readable and typically maintained within a text file, and compiled or machine code which is expressed in binary code read by machines. Source code is more accessible than machine code, as it permits direct modification by its original or subsequent developers. Open source code, by extension, is twice more accessible than compiled software. While the first level of closedness is a constraint of the translation of binary code into electrical signal, the second level of closedness is a constraint of the market. The less common usage of the term ‘closed source’ means fee-based or otherwise proprietary access to the use of particular software that does not also convey access to its human-readable non-compiled code (Fuller & Fuller, 2014).
It seems fairly evident that a quality of openness extends to the source code, and only by extension to the whole software. Less clear is whether the relationship between source and machine code (together becoming a software) is metonymy or synecdoche. In other words, is the desire for access driven by a want of the thing, or by a want to revise and thereby refine the thing. This fundamental uncertainty is at the heart of the need for disambiguation of the related concept of free(dom), and the frequent need for qualification through the use of accumulated descriptors free libre and open source (F/LOS).
I find it revealing that in Fuller’s entry for source code within Software Studies: A Lexicon, while extolling the capacity for open source code to enable artistic and creative modes of self-expression within code, Fuller is also quick to include as an example of this creativity the International Obfuscated C Code Contest as well as other instances of software creators using deliberately exclusionary or quasi-proprietary practices in order to complicate various kinds of access. These may include use without attribution or use without sufficient comprehension of how things work. Whether a knowing wink to another cognoscente, or an act of ironic expression, these means of boundary enforcement serve as another form of closedness and exclusionary practice that reifies a kind of artificial scarcity similarly created within the marketplace as perpetuated by licensing and other kinds of access policing.
Beyond these: a placeholder for a future entry within Digital Keywords, the pedagogy-focused entry within Digital Pedagogy in the Humanities, and the ‘source code’ entry within Software Studies: A Lexicon, mention of ‘open’ is notably lacking. Several works within the vein of Keywords address one or more related concepts, such as ‘access/accessibility’, for instance. But effort to tease apart varying and nuanced applications of the word ‘open’ is insufficient. ‘Open’ as a descriptive concept is problematic in part because of the broad utility of the word. Openness exists within an ‘Access’ semantic frame and therefore strengthens an argument that much thinking and work around ‘open’ can be understood as a response to markets and class structures. While some degree of access might be mitigated through ‘open’ strategies and constructs, other forms of access control, such as prestige and obfuscation, arise to put new or similar constraints upon the commons.