Digital Studies

How do we know what Shakespeare’s plays sounded like in his time, or Sappho’s verses, or the tales of ancient Sumer? As they were written in phonetic scripts, modern historical linguists have largely been able to reconstruct the sounds of these works. Written Chinese has always been a logographic and not a phonetic script, and with the rapid pace of phonetic
variation and change, many of the euphonic patterns in ancient Chinese texts of ritual and history have been lost for millennia. While very general categories of rhyme and correlations between characters based on ancient rhyming poetry have been proposed by Chinese scholars throughout the ages, until developments in Western historical linguistics were applied to Chinese over the past century, the sounds of this ancient language remained obscure. However, thanks to modern advances in computer database technology, digital texts and digital tools, a wide variety of phonological data for ancient Chinese (including several recently-developed systems for reconstructed pronunciations) can now be employed to provide empirical documentation and analysis of the lost euphony and phonorhetorical structures in these ancient texts for the first time.

    In this study I utilize a tripartite framework for philological inquiry, grounded in the equal consideration of semantics, metrics and acoustics. In general, over the past two millennia, most Chinese philological studies have focused upon detailed exegeses of the semantics of a word, passage or text. Metrical features and sentence prosody have also received some attention, as various forms of literary expression in Chinese have been governed by conventions of style and form; this is particularly true of poetry, but also of patterned and parallel prose. This study argues that analyses of the phonetic patterns in a text should also play a significant role in any significant philological study, as it is often in the pairing of acoustic devices with metric and semantic structures that the true breadth, depth and beauty of literary expression can be felt most acutely. This framework represents a methodological shift in Chinese philology: until recently it was extremely difficult to accurately assess phonetic and acoustic structures in early Chinese texts; this was particularly true for compositions from the distant past. However, thanks to modern technology and recent advances in the field of Chinese phonology, it is now possible for any scholar to efficiently evaluate the acoustic structures of any Chinese text with as much accuracy as the aggregate of available phonological evidence can provide and thereby gain a more complete understanding of its acoustic constitution, its aesthetic and performative features, and the more subtle aspects of literary artistry which informed its composition and transmission.

    The phonological foundation of this study has been facilitated by a digital suite of lexical tools which I designed and which are the first method by which the hurdles to large-scale Chinese lexical spadework in the service of phonological analysis can efficiently be overcome: The Digital Etymological Dictionary of Old Chinese (available online at edoc.uchicago.edu ), via which one can programmatically retrieve a wide range of phonological data, from both ancient and modern resources, for every character in any Chinese text. For this study, I used it to compile and evaluate proposed Old Chinese pronunciations for each graph in three of the earliest corpora of Chinese narrative texts: inscriptions longer than fifty graphs preserved on bronze vessels dating to the Western Zhou dynasty (1045-771 B.C.E.), the ten chapters of the Classic of Documents《尚書》 which scholars now generally believe were likely originally composed during the
Western Zhou, and speeches of over one hundred graphs preserved in the Zuo Commentary to the Spring and Autumn Annals《春秋左傳》. From these results, I chose four representative inscriptions from the Western Zhou inscriptional corpus, two representative chapters from the Classic of Documents, and three representative speeches from the Zuo Commentary as the basis for the analyses in chapters two through four. In these case studies, complete charts of each text (including a full transcription in Chinese, an Old Chinese phonological reconstruction for each graph, and an English translation) are provided, followed by detailed evaluations of the euphonic patterns and phonorhetorical devices employed within each text. The concluding chapter presents a brief overview of the main types of euphonic patterns and phonorhetorical devices evidenced within each corpus followed by general remarks on the euphonic and phonorhetorical characteristics common to all three corpora, and finds that there are demonstrable commonalities yet each corpus exhibits a unique range of euphonic and phonorhetorical devices which distinguishes it from the others, and from other early Chinese literary genres.

The goal of epigraphic study is to observe developments in the forms of signs and letters over time and in different scribal hands. With enough information, the idea is that one can posit the date of a text based on the forms of the signs or letters. If precise date ranges of texts are not possible, one hopes to establish a relative chronology, at least. In addition, one can attempt to establish scribal styles: styles by location, scribal school, genre, etc. The practice of epigraphy usually involves creating charts that take the form of line drawings of signs or letters. In a pre-digital context, a researcher would carefully sketch and ink letter forms, arranging them in columns for comparison. These charts are very helpful but very limited. In many cases, it is not clear to the reader if a given example is taken directly from a text or is a generalized form. Also, it is typically too cumbersome for the researcher to note the exact source of each sign. In a digital world now, it is time for epigraphy to evolve. This paper presents a radical and powerful new strategy for creating letter charts from an integrated database environment that includes textual, lexicographic, and archaeological data. The database environment provides complete integration between photographs, line drawings, text editions, lexical entries, find spots, and many other categories of analysis. The database environment that I will present is called OCHRE, the Online Cultural and Historical Research Environment. OCHRE can be accessed for free from any computer with an internet connection. So, any user anywhere in the world will be able to query text corpora and genres and generate letter charts.

I will use the alphabetic texts from Ras Shamra-Ugarit as my example corpus. The majority of the alphabetic Ugaritic texts dates to c. 1200 BCE. As such, diachronic epigraphic studies are not possible. However, one observes several distinct epigraphic styles that seem to correspond grosso modo to textual genres. The Ugaritic letter forms are produced with a fine and careful hand in the mythological texts (for the most part). The economic texts are characterized by a broad heavy hand in many cases. Also, the Ugaritic alphabetic letters exhibit morphological variation in their basic forms.

The paper presents the database environment, the data model, and the strategy for working on epigraphy in a database. The result is a dynamically updatable epigraphic chart of Ugaritic letters, some based on photographs of tablets, some on hand drawings. All of this data is generated from a database that is richly described with properties and links and can be accessed from any computer. The entire process is generic enough to be applied to any text corpora, whether alphabetic or syllabic, ancient or modern.

Capturing, organizing, and displaying accurate and useful images is a challenge for any research project. This paper will present a field-tested comprehensive system for digital imaging for archaeological and philological research projects, from image capture, to integration, to presentation and preservation. Powerful methods for capturing field and object photos will be presented, including recently field-tested approaches to aerial photography for photogrammetry and reflectance transference imaging for low relief objects like inscriptions and seal impressions.

Creative and powerful approaches to integrating project data and digital images will be presented in detail, with examples from active research projects such as the Neubauer Expedition to Zincirli and the Persepolis Fortification Archive Project. In these examples, project data and images are integrated in a comprehensive hierarchical item-based database system called OCHRE, the Online Cultural and Historical Research Environment, where thousands of images, plans, artifact descriptions, analyses, and notes are organized and interconnected through various strategies.

Within this highly atomized and systematically organized approach, project data can be easily retrieved and displayed together online as a digital publication. This approach to data organization also allows for easy preservation of data, a consideration which has already become a requirement by most granting institutions.

for their help with proofreading. In addition, we thank the Brill editors, and especially the editors of the Digital Biblical Studies series, Claire Clivaz and David Hamidović, who patiently supported us in this wonderful project. Finally, the co-editors, Vanessa Bigot Juloux, Amy Rebecca Gansell, and Alessandro di Ludovico, sincerely thank the authors for their close collaboration and for having agreed to join us in this adventure toward the advancement of the humanities for both educational and scientific purposes. More than anything else, this project would never have been successful without the strong support of our families (including Vanessa's numerous cats and dogs), friends, and peers, of whom we are deeply appreciative.

Twenty-five hundred years ago, the "paperwork" of the Achaemenid Empire was recorded on clay tablets. In 1933, archaeologists from the University of Chicago's Oriental Institute (OI) found tens of thousands of these tablets and fragments during the excavation of Persepolis. Many of these tablets have been painstakingly photographed and annotated by expert cuneiformists, and now provide a rich dataset consisting of over 5,000 annotated tablet images and 100,000 cuneiform sign bounding boxes. We leverage this dataset to develop DeepScribe, a modular computer vision pipeline capable of localizing cuneiform signs and providing suggestions for the identity of each sign. We investigate the difficulty of learning subtasks relevant to cuneiform tablet transcription on ground-truth data, finding that a RetinaNet object detector can achieve a localization mAP of 0.78 and a ResNet classifier can achieve a top-5 sign classification accuracy of 0.89. The end-to-end pipeline achieves a top-5 classification accuracy of 0.80. As part of the classification module, DeepScribe groups cuneiform signs into morphological clusters. We consider how this automatic clustering approach differs from the organization of standard, printed sign lists and what we may learn from it. These components, trained individually, are sufficient to produce a system that can analyze photos of cuneiform tablets from the Achaemenid period and provide useful transliteration suggestions to researchers. We evaluate the model's end-to-end performance on locating and classifying signs, providing a roadmap to a linguistically-aware transliteration system, then consider the model's potential utility when applied to other periods of cuneiform writing.