AWRC in Lockdown

Despite not being able to carry out many of the social and networking events for which the Ancient World Research Cluster is known, our members have been active as ever during Trinity Tem and the Long Vacation and have found imaginative ways to develop their various research projects.

You can find below a selection of the many projects sponsored by the Lorne Thyssen Fund for Ancient World topics and discover how our members are coping in these challenging times. 

Mr Michael Macdonald (HF)

Just before lockdown struck Michael Macdonald and Ali Al-Manaser, from the Ministry of Tourism and Antiquities of Jordan, finished the 2020 season of the Badia Epigraphic Survey (1st-14th March) in the desert of north-eastern Jordan. The team comprised two other members, Dr Orhan Elmaz, Senior Lecturer in Arabic at St Andrews University, and Abdallah Hatlani a Kuwaiti PhD student at Leiden University whose thesis is on early Arabic inscriptions. Michael kindly shares the highlights of the season with us:

"The area visited this season was at the point where the ḥarrah, or basalt desert of broken up lava-flows, ends and the underlying ḥamād or limestone desert becomes visible again (see photo). This area is particularly interesting because we know from the "Safaitic" graffiti left by ancient nomads some 2000 years ago that this was where they gathered at the end of the dry season (roughly mid June to early October) waiting for the first rains of the year which would quickly cause herbage to sprout up in the limestone desert. We surveyed parts of this border-land in 2018, but this year we were especially privileged to be able to go right up to the area next to the Syrian border and, accompanied by soldiers from the Jordanian army, even into no-man's-land. This permission was negotiated with the army by Dr Ali Al-Manaser of the Ministry of Tourism and Antiquities and allowed us to survey areas which had never been visited by archaeologists or epigraphists before. We are most grateful to the Jordanian army and the Ministry of Tourism and Antiquities for making this possible.

The results were spectacular and we recorded some 2000 inscriptions in the two weeks of the survey. We also discovered an ancient burial cairn which was intact and had not been plundered by the numerous seekers after (supposed) buried gold — a fate which has befallen large numbers of ancient and modern burials in other parts of the ḥarrah. This ancient burial cairn could be identified by the large numbers of Safaitic inscriptions on it saying that the author mourned and/or helped build the cairn for the deceased. For the first time we found here a text by woman saying that she helped build the cairn for her maternal uncle.

We examined the whole area and systematically surveyed those parts where the rock was inscribable and those where the types of basalt were mixed [...] We recorded all forms of carving: the ancient (Safaitic) graffiti and rock drawings, early and mediaeval Arabic graffiti, modern Arabic graffiti and drawings — for the Bedouin have become literate again thanks to mobile phones and are using their skills in the same way as their ancient predecessors — and tribal marks (wusūm), which were often used as a form of identification before literacy became widespread, on graves or as a way of saying "Kilroy was here". There were some magnificent ancient drawings of horsemen and camels, including a Bactrian camel— an import from Central Asia probably captured from a caravan on the Silk Road — (see photo) as well as modern ones of trucks, and pictures of parties with people dancing and playing musical instruments, battles and hunts. We also discovered several desert mosques, areas enclosed by low walls or simply lines of stones with a qibla, or small apse indicating the direction of prayer.

Every find was photographed with cameras which recorded the GPS location and these locations are at present being mapped by Orhan Elmaz. For inscriptions and drawings, the find was recorded within its topographical context and its epigraphical context (i.e. with the inscriptions around it), and then with general photographs and details. Many inscriptions were partially buried in the mud and so had to be dug out and then washed and scrubbed and left to dry before we could photograph them. All in all, we took over 37,000 photographs in 14 days of fieldwork producing a massive post-survey task for which self-isolation is ideally suited!

We had the full range of weathers: some days were very hot, others extremely cold with gales blowing and sometimes driving rain. Fortunately, except for one day when the rain was continuous, it did not continue for very long at a time and the wind soon dried the stones. We had taken 30 umbrellas to shade the stones when we photographed them and of these only 2 survived the fortnight!

The Badia Epigraphic Survey is profoundly grateful to the Lorne Thyssen Research Fund for Ancient World Topics for its very generous grant which has made this year's fieldwork possible. We hope that you will be pleased with the results.

Mr Tim Moller (GS)

Tim has been working hard and enthusiastically on his Master's thesis, which examines elephant and ivory extraction by the Ptolemies in Nubia in the late third century BC, and is expanding that research into an investigation of the role of the elephant in the Hellenistic world. This work will support an eventual DPhil application for 2021. His fieldwork in the Nile valley, from Khartoum to Cairo in March 2020 was supported by the AWRC and was cut short due to the coronavirus outbreak. He is planning further work, also sponsored by the AWRC, in the Autumn.

Do you have 5 minutes? Listen to his Tim's research here.

Dr Thea Sommerschield (GS)

Thea has just defended her dissertation, entitled "Identities through boundaries: a study of patterns of practice in the communities of Western Sicily in the late Archaic and Classical period". She has travelled extensively during her DPhil work, supported by the AWRC, among others. During the lockdown, she has been working on various projects related to her interest in Computational Humanities: Pythia, the first Deep Learning text restoration model for Greek Epigraphy, and ISicDef, a digital relational database of Sicilian curse tablets. Her current and future research projects explore how Machine Learning can enable large-scale, SOTA interpretations of the textual evidence from the Ancient World.

In November 2019, sponsored by the AWRC, she attended the "2019 Conference on Empirical Methods in Natural Language Processing and 9th International Joint Conference on Natural Language Processing" (see photo). Here her report:

"I presented a research project applying Artificial Intelligence to the field of Ancient History. The project's name is PYTHIA: it is the first epigraphic restoration model that recovers missing characters from a damaged text input using deep neural networks. This work is co-authored by myself, Yannis Assael (Google DeepMind) and my DPhil supervisor Professor Jonathan Prag (University of Oxford).

Ancient History relies on disciplines such as Epigraphy, the study of ancient inscribed texts (“inscriptions”), for evidence of the recorded past. However, inscriptions are often damaged over the centuries, and illegible parts of the text must be restored by specialists, known as epigraphists. Epigraphic restoration is a complex and time consuming task, and epigraphists would greatly benefit from an assistive system that can furnish a more complete and systematically ranked list of possible solutions, improving the accuracy and speed of the textual restoration pipeline. Bringing together the disciplines of Ancient History and Deep Learning, PYTHIA offers a fully automated aid to the text restoration task, providing ancient historians with multiple textual restorations, as well as the confidence level for each hypothesis. EMNLP-IJCNLP 2019 was held at the Asia World Expo in Hong Kong.

The EMNLP-IJCNLP 2019 proceedings, including our published paper, are available in the ACL anthology (Google Scholar H5-index 88, ranked #2 in the top Computational Linguistics publications). PYTHIA was paper #1020 in Session 12 “Information Extraction, Text Mining and NLP Applications, Social Media and Computational Social Science, Sentiment Analysis and Argument Mining”.

The research was extremely well received, and has attracted the attention of Google and Facebook research scientists, university researchers (including Cambridge, Oxford and the Turing Institute), and technology startups. I also networked with researchers who are very interested in collaborating on future research projects applying Machine Learning to the textual cultures of the Ancient World. While attending the talks and poster sessions, I was particularly interested in the applicability of the new state of the art BERT language model to my next research projects on digital epigraphy. I also attended the three keynotes, given by Meeyoung Cha, Kyunghyun Cho, and Noam Slonim, which I found particularly interesting with regard to ongoing interdisciplinary projects and the state of the art in Machine Learning and Artificial Intelligence. In fact, each session I attended gave me much to think about the potential for applying certain techniques and models to researching the evidence from the Ancient World and the Humanities.

To conclude, attending and presenting at EMNLP 2019 has offered me the unprecedented opportunity to engage and exchange ideas with specialists in the field. I have also been well advised on how best to proceed applying the PYTHIA model to Latin and to other low-resource languages such as Punic. I wish to express my sincere gratitude to the Lorne Thyssen Fund, for having made this research activity possible."

Dr Elizabeth Tucker (SF)

Elizabeth has been busy during lockdown working on a edition, translation and commentary of Book 11 of the Paippalāda Atharvaveda. Some results of this research were presented last year at an international workshop in Dubrovnik, for which she received the support of the Lorne Thyssen Fund for Ancient World Topics. These are her words:

"Last summer I received support from the Lorne Thyssen Fund for Research in Ancient World topics to attend the 7th International Vedic Workshop, held at the Inter-University Centre/ Centre for Advanced Academic Studies of Zagreb University, in Dubrovnik, 19 -24 August 2019. This Vedic Workshop provided an opportunity for me to present some results from my current research on the Atharvaveda of the Paippalādins, a brahminical school of Ancient India.

The Atharvaveda, the Fourth Veda, is sometimes called “the Veda of magic and spells”, because its content differs from that of the three Vedas that were employed in the solemn ritual. It has been transmitted in two recensions, one of which, the Atharvaveda of the Śaunaka school, has been studied by both Indian and Western scholars since the mid-19th Century. The Paippalāda recension was known only from one very corrupt birchbark manuscript written in the Śāradā script (see photo), which was discovered in Kashmir in 1874. However, during the1950s and 1990s Paippalāda palm-leaf manuscripts written in the Oriya script (see photo) were discovered in the Indian State of Odisha (Orissa) and these are at last providing a comprehensible text for the Paippalāda Atharvaveda, which contains a considerable number of previously unknown Vedic mantras and longer poems.

In the course of editing and translating a composition from its 11th book, I noticed that it contained two interesting Vedic Sanskrit words, which are not recorded in the lexica used by scholars. One is a feminine noun gṛtsī- which is clearly related to the Vedic adjective gṛ́tsa- ‘wise’. The Paippalāda 11 context shows that it must mean ‘wise woman, witch’. The other is a masculine noun ṛṣṇu-, which occurs in a parallel context later in the same poem and which therefore refers to a male ‘sorcerer’ of some sort. The Atharvaveda contains other words for sorcerers, but my paper suggested that there is linguistic evidence to connect ṛṣṇu- with the well-known Vedic and Sanskrit noun ṛ́ṣi- for which the stock translation is ‘sage’, but which also in the Rigveda means ‘a composer of verses or hymns’. It can be envisaged that both related nouns referred to a person who knew or composed powerful mantras or verbal formulae (an old etymology suggested by H. Grassmann pointed to a literal meaning ‘fluent’ for ṛ́ṣi-); but that, whereas the ṛ́ṣi- was associated with using his skills for good ritual purposes, the ṛṣṇu- became associated with bad or hostile uses and so in the Atharvaveda the word designated a ‘composer of hostile spells’.

These are examples of material which is of particular interest to philologists that the Paippalāda Atharvaveda is turning up. My work of editing, translating and commenting on the rest of Book 11 on the basis of photos of the manuscripts is continuing under lockdown."

Dr Susan Walker (EF)

Susan tells us about the exciting research that she and Dr Marshall Woodworth are carrying out at Steane Park, Northamptonshire, and the question they are trying to answer using advanced lab techniques. This is her report:

"Last year I applied for a grant from the Lorne Thyssen Research Fund to support residue analysis of potsherds recovered in the 2013 excavations at Steane Park, Northamptonshire. Completion of the final report has been dogged by difficulties, not least the pandemic, and I am grateful to have the grant offer extended for the next academic year. I very much hope that Dr Marshall Woodworth, who carried out a pioneering and successful analysis of the pewter platters from the excavation, will have access to the Research Laboratory for Art History and Archaeology to complete his work, using a newly developed technique.

The research question is a simple one requiring a yes/no answer: do the residues on the ceramics match those found on the pewter platters? 

The answer will tell us if individual diners were consuming the same food as was offered on the buried platters. Preliminary analyses were consistent with a ruminant animal: cow, sheep or goat. The animal bones present on site indicate that any of these is possible, but cow bones are dominant. The careful balancing of the platters on uneven ground indicate that the butchered animal had been cooked in liquid. Slow boiling of raw meat in a cauldron suspended or propped over an open fire is a well attested culinary practice in late Roman Britain.

Residue analysis may also reveal the purpose of the enormous globular storage jars, locally made and very prominent in the ceramic record from the site, though it probably won’t be clear, unless there are layers of residue, whether this was a primary or later use of the jars, which could have been reused as cooking vessels.

The completed excavation report indicates that this part of the site was used as a kitchen and dining area. The deliberate deposition of the pewter platters (see photo) within a partly ruined Roman building suggests a feast to mark the end of the use of Steane as a religious site in the late fourth century AD. The feast may even have been organized by the managers of the next phase of activity on this part of the site. The hillside was then quarried for ironstone, from which nuggets of iron ore could be extracted for smelting. Finds of bronze belt accessories, a burial on-site and respect for nearby earlier depositions suggest that this was a managed activity happening at the turn of the fourth and fifth centuries AD, with no build-up of soil between the evidence for feasting and the quarry-holes."