Bias in metadata 2: The Influence of polyvocality on the life-cycle of the GLAM objects

This blog post is the second of three blog posts, published by the Dutch Heritage Network Preservation Watch, which monitors technological developments as relevant to the GLAM sphere.

The blog posts are written by Ryan Brate, who is a PhD candidate within DHLab of the Koninklijke Nederlandse Akademie van Wetenschappen. Read the interview with Ryan and the first blog of the series on Kennisplatform Preservation.

Many western museum collections have their origins in the colonial period and often contain complex and charged content and themes in their respective collections. The content of such collections may be centuries old, and the objects themselves carefully preserved, but our retrospective interpretation of them is very much subject to change and re-evaluation. Increasingly, museum professionals and institutions are making revisions to their collections and object descriptions, in the interests of better representing previously marginalised viewpoints. In this blog post, we explore how the increased recognition of previously marginalised viewpoints, and the importance attributed to them, is being played out in GLAM institutions in relation to the life-cycle of collection objects. That is, how museums are handling the competing polyvocal narratives that can be attributed to terminology, themes and objects?

GLAM decolonisation

Decolonisation is the process of making changes to exhibit portrayals in such a way that broadens the perspectives portrayed beyond those of some prevailing narrative. Decolonisation represents perhaps the most significant ongoing movement related to object lifecycle in GLAM institutions as a consequence of increased recognition of previously marginalised voices.

The decolonization of the European powers and the establishment of the independence of the colonised, is very recent history. Its effects continue to bear fruit in the present day in discussions of social justice, equality, ethnicity and cultural values. Colonial history is undeniably and understandably contested. In recognition of this, there is an increasing movement towards decolonization by and of academic institutions: to identify, re-evaluate and critique the narratives propagated by (for example) objects and language rooted in colonial viewpoints. Such decolonization efforts are being conducted by groups within GLAM institutions, but also by universities and student groups (e.g., Cambridge University Decolonisation Working Group, UvA Amsterdam Centre of European Studies). This coincides with notable instances of ongoing societal discussions and actions in relation to how European colonial powers deal with their past, e.g., the legacy of Cecil Rhodes in the UK or the legacy of King Leopold of Belgium with respect to the Democratic Republic of Congo.

GLAM institutions are central to any such effort, as the value of their collections stems largely from what we are able to infer from them about our shared history. The pieces of a collection, the way they are assembled and described, the terminology used, all contribute to impressing on the observer some set facts or messaging. This is particularly compounded by the fact that Western museum collections may originate from colonial times and so their object depictions and descriptions may be very much of the time. In the Netherlands there have been a number of notable and oft-highly publicised initiatives by GLAM institutions and others, as related to ongoing de-colonisation. For example:

• In 2019, Amsterdam Museum issued a statement that the institution will remove all references to Gouden Eeuw (Golden Age) in its galleries. In doing so, the museum noted the negative aspects of 17th Century Dutch Colonial activity were being masked by the phrase, in being used as synonymous for the 17th Century;

• The previous year, The Mauritshuis, removed a bust of its namesake and for whom the Mauritshuis was a former residence, Johan Maurits van Nassau-Siegen. Governor of the Dutch West India Company from 1636 to 1644, Johan Mauritz oversaw a colony in Brazil engaged in the use of slave labour;.

• The Tropenmuseum’s Words Matter publication, which attempts to offer guidance to similar such institutions as to why certain terminology is problematic and offers suggestions of more appropriate alternatives in certain contexts.

Of course none of these initiatives were met with universal praise. In response to Amsterdam Museum’s move of removing mention of the phrase Gouden Eeuw, Prime Minister Mark Rutte is reported to have called the move “nonsense”, with his Education Minister Arie Slob responding, “I’m a bit tired of discussions about one term.” However, such resistance to changing norms is to be expected and perhaps is no bad thing at all, in the context of wanting to actively engage greater discussion and awareness on particular topics.

Out with the old?

It is noteworthy that each of the aforementioned initiatives are ones of obsolescence. Where in recognition of previously marginalised viewpoints, objects or terminology are being retired and removed from general or particular use. Recognition of problematic objects and content is one thing, but how best manage the implications of its life-cycle connotations in the present day, is another openly-debated discussion. Any action or inaction being open to criticism and unlikely to satisfy all.

Anecdotally, wholesale removal of offending language or items often seems to draw criticism of being tantamount to erasing aspects of history. For example, in the case of the removed Mauritshuis bust, De Telegraaf ran a piece critical of the action with the title, Stop de vervalsing (Stop the falsification). In the UK, the removal and vandalism of a statue of slave trader Edward Couldon, prompted then Prime Minister Boris Johnson to comment that it was wrong for people to "go round seeking retrospectively to change our history". Such arguments may sometimes be paper-shields masking other sentiments, but there is also arguably some merit to the position. Leaving objects in place risks perpetuating biased accounts of history, but their removal may be a pathway to lessening historical understanding and awareness. There are several such examples of problematic objects, put on display with additional context in order to facilitate discourse. For example, the recent retiring of, “De Gouden Koets” (The Golden Coach), of the Dutch Royal Family whose imagery is suggestive of the subjugation of colonised peoples. This is currently housed in the Amsterdam Museum as a short-term loan exhibit.

In terms of language use, there are also reasonable arguments which support the removal or restriction of certain terms. For example, the Words Matter publication includes a compelling short essay entitled, Language cannot be “cleaned-up”, by Esther Peeren. In the essay, referencing the work of Russian thinker Mikhail Bakhtin, it is postulated that language has different implications and weight depending on the socio-cultural position of the user. And hence, that “it matters who says what and in what context, and it can therefore be completely logical to restrict the use of certain words to particular groups of speakers”. Indeed, it is easy to think of particular race-related terms that have become recognised as highly derogatory and which there is little to be gained from their casual use in any setting. However, the same publication highlights that sometimes a term is highly context dependent as to whether it is deemed problematic. For example, the word exotic may be deemed (by some) to have an inappropriately sensuous connotation when applied to people, but is otherwise perfectly innocuous when applied to plants. There are several such examples provided: the word primitive when applied to a people group or culture may imply lesser and that another group is better. However, primitivism as an art form or primitive as applied to an early stage of some development are fine. The connotations of the word, marron are described to vary not only in the context it used, but varies drastically in its connotations as recognised by different affected groups. Marron is said to be positively viewed as a term in Suriname owing to its association with colonial resistance in Suriname. Context dependence adds a new dimension and challenge to this discussion: wholesale removal and replacement of terms in digital archives may be straightforward, but it ignores context and polysemy.

To summarise, there seems to be a broad and increasing consensus among heritage professionals and institutions that decolonisation is a real issue that needs to be tackled. There of course exists no playbook with easy answers as to how it should be done. It remains an open and item by item discussion. Interestingly though, there is something of a playbook regarding the communication of decolonization strategies. The Museums Association, a professional membership organisation for and by heritage professionals, has produced a guidance document with use case examples of strategy communication.

Enrichment as a strategy

Contextualisation is key in presenting sensitive information to users. Data enrichment is about adding supplementary information to object descriptions, to help provide further contextual understanding to the reader. The publication Echoes: Practising Decoloniality in Museums: A guide with Global Examples gives several examples of contextualisation by GLAM institutions, either as new exhibits or amendments to existing exhibits. For example, in response to social media criticism from the #DecolonizeTheMuseum activist group, the Tropenmuseum responded by facilitating the placing of additional explanatory panels of text next to exhibits by the group. These panels detailed colonial exploitation in relation to the exhibits, but also the role of the museum itself in acquiring objects. Such amendments, of course, are ultimately limited in their scope and such an approach would not be feasible for the enormous quantity available in the digital collections setting.

Digital collections present a particular opportunity for contextualisation of items, given that enrichment can be done to some extent en-masse and somewhat dynamically. However, (semi-)automatic enrichment is not trivial. When working with text collections, the very nature of language presents two main problems: ambiguity - that is one term can mean multiple things, and polysemy - that is that the same idea can be expressed in multiple ways. Due to ambiguity in language, we cannot simply take a list of offensive terms and replace them with non-offensive terms, as most terms are only offensive in certain contexts. On the other hand, language use is productive and creative, thus it is impossible to create a list of all potential synonyms in all potential contexts that people could have used to express an offensive idea. However, this does not mean that data providers cannot do anything - terms can be flagged as potentially offensive, with an explanation. The Dutch National Archives already provide users an option to notify them of potentially problematic terminology - this could be supported by AI techniques that are being developed for offensive language [1] [2] [3]. Another interesting research area is the semantic web, that aims to capture and represent knowledge in a structured manner, as is discussed in our next blog post.

The complete series by Ryan Brate:
blog 1: Monitoring advances in the field of AI, with an emphasis on bias
blog 2: The Influence of polyvocality on the life-cycle of the GLAM objects
blog 3: Adding (polyvocal) context to semantic web representations
Interview with Ryan Brate