“The density of types and the dignity of the fragment. A website approach to archaeological typology” by G. Buccellati and M. Kelly-Buccellati (1) is a contribution to the rapidly growing literature on digital approaches to archaeological data management, expertly showcasing the significant theoretical and epistemological impetus of such work. The authors offer a conceptually lucid discussion of key concepts in archaeological ordering practices surrounding the longstanding tension between so-called ‘etic’ and ‘emic’ approaches, thereby providing a thorough systematic of how to think through sameness and difference in the context of voluminous digital archaeological data. 

As a point of departure, the authors reconsider the relationship between archaeological fragments – spatiotemporally bounded artefacts and features – and their larger meaning-giving totality as the primary locus of archaeological knowledge. Typology can then be said to serve this overriding quest to resolve the conflict between parts and wholes, as the parts themselves are never sufficient to render the whole but the whole remains elusive without reference to the parts. Buccellati and Kelly-Buccellati here make an interesting point about the importance to register the globality of the archaeological record – that is, literally everything encountered in the soil – without making any prior choices as to what supposedly matters and what not. The distinctiveness of the archaeological enterprise, according to them, indeed consists of the circumstance that merely disconnected fragments come to the attention of archaeologists and the only objective data that can be attained, because of this, are about the situated location of fragments in the ground and their relation to other fragments – what they call ‘emplacement’. This, we would add, includes the relationship of fragments with human observers and the employed methods of excavation as observation. As the authors say: “[i]t is in this sense that the fragments are natively digital: they are atoms that do not cohere into a typological whole”.

The systematic exploration of how the so recovered fragments may be re-articulated is then essentially the goal of archaeological categorization and typology but these practices can only ever be successful if the whole context of original ‘emplacement’ is carefully taken into consideration. This reconstruction of the fundamental epistemological situation archaeology finds itself in leads the authors to a general rejection of ‘more’ vs. ‘less’ objective or even subjective ordering practices as such qualifications tend to miss the point. What matters is to enable the flexible and scalable confrontation of isolated archaeological fragments, to do experiment with and test different part-whole relations and their possible knowledge contributions. It is no coincidence that the authors insist on a dynamical approach to ordering practices and type-thinking in archaeology here, which in many ways comes often very close to the general conceptual orientation philosopher Stephen C. Pepper (2) has called ‘organicism’ – a preoccupation of resolving the tension between heterogeneous fragments and coherent wholes without losing sight of the specificity of each single fragment. In the view of organicist thinkers, and the authors seem to share this recognition, to take complexity seriously means to centre the dialectics between fragments and wholes in their entirety. This notion is directly reflected in the authors’ interesting definition of ‘big data’ in archaeology as a multi-layered and multi-referential system of organizing the totality of observations of emplacement (the Global record).

Based on this broader exposition, Buccellati and Kelly-Buccellati make some perceptive and noteworthy observations vis-à-vis the aforementioned emic-etic distinction that has caused so much archaeological confusion and debate (3–6). To begin with, emic and etic designate different systemic logics of organizing observable sameness and difference. Emic systems are closed and foreground the idea of the roster, they recognize only a limited set of types whose identity depends on relative differences. Etic systems, on the other hand, are in principle open (and even open-ended) and rely on the notion of the lexicon; they enlist a principally endless repertoire of traits, types and sub-types (classes and sub-classes may be added to this list of course). Difference in etic systems is moreover defined according to some general standards that appear to eclipse the standards of the system itself. Etic systems therefore tend to advocate supposedly universal principles of how to establish similarity vs. difference, although, in reality, there is substantial debate as to what these principles may be or whether such endeavour is a useful undertaking. In the wild, both etic and emic systems of ordering and categorization are of course encountered in the plural but etic systems deploy external standards of order while emic systems operate via internal standards. An interesting observation by the authors in this context is that archaeological reasoning in relation to sameness and difference is almost never either exclusively etic or exclusively emic. The simple reason is that any grouping of fragments according to technological (means/modes of production) or functional considerations (use-wear, tool design, relation between form and function) based on empirical evidence is typically already infused by emic standards. The classic example from the analysis of archaeological pottery is ware groups, which reference the nexus of technological know-how and concrete practices, and which rely, in a given context, on internal, relative differentiations between the respective observed practices. Yet ignoring these distinctions would sideline significant knowledge on the past.

These discussions are refreshing as they may indicate that ordering practices – when considered as an end in themselves – misconstrue the archaeological process as static and so advocate for categories, classes, and types to be carved out before any serious analysis can begin. It could in fact be argued that in doing so, they merely construct a new closed system, then emic by definition. Buccellati and Kelly-Buccellati propose an alternative without discarding the intuition that ordering archaeological materials is conditional to the inferential and knowledge-production process: they propose that typologies should be treated as arguments. Moreover, the sort of argument they have in mind is to a lesser extent ‘formal-logical’ but instead emphatically ‘dialogical’ in nature, as such argumentative form helps to combat the inherent static-ness of ordering practices the authors criticize, and so discloses a radically dynamic approach to the undertaking of fragment-whole matching. The organicist inclination to preserve ‘the dignity of fragments’ while working towards their resolution in attendant wholes and sub-wholes further gives rise to the idea that such ‘native digital fragments’ must be brought into systematic conversation with one another, acknowledging the involved complexity. To this end, the authors frame ordering work and typo-praxis as a ‘digital discourse’ and ask what the conditions and possibilities for such discourse are and how it can be facilitated. It is here that they put forward the idea that the webpage may provide an ideal epistemic model system to promote the preservation of emplaced archaeological fragments while simultaneously promoting multistranded and multi-context explorations of fragment coherence and articulation. The website enables unique forms of exploration and engagement with data and new arguments escaping the fixity of the analogue-printed which dominates current archaeological practice. Similar experiences were for example made in the context of Gardin’s ‘logicism’, leading to broadly comparable attempts to overcome the analogue with more dynamic, HTML/web-based forms of data presentation, exploration and discussion (7, 8).

As such, Buccellati and Kelly-Buccellati table a range of fresh arguments for re-thinking typology beyond and with text at the same time, to enable ‘dynamic reading’ of fragment-whole relationships in an increasingly digital world. Their proposal comes thereby close to what has been termed ‘deep mapping’ in the context of critical cartographies and other spatially-inclined scholarship in the Anglophone world (9, 10). Deep maps seek to transcend the epistemological limitations of 2D-representations of spatiality on traditional maps and introduce different layers of informational depth and heterogeneity, which, similarly to the living digital webpage proposed by the authors, can be continuously extended and revised and which may also greatly promote multidisciplinary and team-based research endeavours. In the same spirit as the authors’ ‘digital discourse’, deep mapping draws attention to the knowledge potential of bringing together the heterogeneous, the etic and the emic, and to pay more attention to ‘multiplanar’ and ‘multilinear’ relationships as well as the associated relations of relations. This proposal to deploy types and typology in general as dynamic arguments is linked to the ambition to contribute to and work on the narrativization of the archaeological record without tacit (and often unconscious) conceptual pre-subscription, countering typologies that remain largely in the abstract and so have contributed to the creeping anonymity of the past.

Bibliography

1. Buccellati, G. and Kelly-Buccellati, M. (2023). The Density of Types and the Dignity of the Fragment. A website approach to archaeological typology, Zenodo, 7743834, ver. 4 peer-reviewed and recommended by Peer Community in Archaeology. https://doi.org/10.5281/zenodo.7743834.

2. Pepper, S C. (1972). World hypotheses: a study in evidence, 7. print (Univ. of California Press).

3. Hayden, B. (1984). Are Emic Types Relevant to Archaeology? Ethnohistory 31, 79–92. https://doi.org/10.2307/482057

4. Tostevin, G. B. (2011). An Introduction to the Special Issue: Reduction Sequence, Chaîne Opératoire, and Other Methods: The Epistemologies of Different Approaches to Lithic Analysis. PaleoAnthropology, 293−296. https://www.doi.org/10.4207/PA.2011.ART59

5. Tostevin, G. B. (2013). Seeing lithics: a middle-range theory for testing for cultural transmission in the pleistocene (Oakville, CT: Oxbow Books).

6. Boissinot, P. (2015). Qu’est-ce qu’un fait archéologique? (Éditions EHESS). https://doi.org/10.4000/lectures.19921

7. Gardin, J.-C. and Roux, V. (2004). The Arkeotek Project: a European Network of Knowledge Bases in the Archaeology of Techniques. Archeologia e Calcolatori 15, 25–40.

8. Husi, P. (2022). La céramique médiévale et moderne du bassin de la Loire moyenne, chrono-typologie et transformation des aires culturelles dans la longue durée (6e—17e s.) (FERACF).

9. Bodenhamer, D. J., Corrigan, J. and Harris, T. M. (2015). Deep Maps and Spatial Narratives (Indiana University Press).

10. Gillings, M., Hacigüzeller, P. and Lock, G. R. (2019). Re-mapping archaeology: critical perspectives, alternative mappings (Routledge).

This paper [1] presents an innovative approach to address the prevalent challenge of unbalanced datasets in coin type recognition, shifting the focus from coin class type recognition to coin mint recognition. Despite this shift, the issue of unbalanced data persists. To mitigate this, the authors introduce a method to split larger classes into smaller ones, integrating them into an 'additional class layer'.

Three distinct machine learning (ML) methodologies were employed to identify new possible classes, with one approach utilising unsupervised clustering alongside manual intervention, while the others leverage object detection, and Natural Language Processing (NLP) techniques. However, despite these efforts, overfitting remained a persistent issue, prompting the authors to explore alternative methods such as dataset improvement and Generative Adversarial Networks (GANs).

The paper contributes significantly to the intersection of ML techniques and archaeology, particularly in addressing overfitting challenges. Furthermore, the authors' candid acknowledgment of the limitations of their approaches serves as a valuable resource for researchers encountering similar obstacles.

This study stems from the D4N4 project, aimed at developing a machine learning-based coin recognition model for the extensive "Corpus Nummorum" dataset, comprising over 19,600 coin types and 49,000 coins from various ancient landscapes. Despite encountering challenges with overfitting due to the dataset's imbalance, the authors' exploration of multiple methodologies and transparent documentation of their limitations enriches the academic discourse and provides a foundation for future research in this field.

Reference

[1] Gampe, S. and Tolle, K. (2024). Creating an Additional Class Layer with Machine Learning to counter Overfitting in an Unbalanced Ancient Coin Dataset. Zenodo, 8298077, ver. 4 peer-reviewed and recommended by Peer Community in Archaeology. https://doi.org/10.5281/zenodo.8298077