Tag Archives: Arabic

Read, Hot and Digitized: Nuṣūṣ — A Corpus of Neglected Texts

Read, hot & digitized: Librarians and the digital scholarship they love — In this series, librarians from the UT Libraries Arts, Humanities and Global Studies Engagement Team briefly present, explore and critique existing examples of digital scholarship. Our hope is that these monthly reviews will inspire critical reflection of, and future creative contributions to, the growing fields of digital scholarship.

While digital, machine-readable texts in Arabic are growing in their availability, certain genres of writing and scholarship in Arabic have become more readily accessible than others. Among those more obscure disciplines are Sufism, theology (Muslim and Christian), and philosophy. These tend to be theoretically complex, and even dogmatically challenging, disciplines that are not as well represented in North American Islamic Studies programs as literature or Qur’anic studies. The Nuṣūṣ corpus––a project led by Antonio Musto––seeks to fill in some of the desiderata by putting more texts from these essential disciplines up on the Internet for researchers to use.

A project that began with an almost exclusive focus on Sufism, Nuṣūṣ has expanded to include works from a variety of complex disciplines of Arabic-language scholarship produced by Muslims and Christians. The corpus currently contains 61 machine-readable texts, with plans to add more and to make the text files available for download. Differing from other, larger corpora of Islamicate[1] disciplines, Nuṣūṣ provides the bibliographic information for the modern editions from which these digitized texts are derived. This is not only a responsible move, but a useful one for researchers: modern editions of historic texts can differ greatly; comparing modern editors’ approaches to the text and their choices that affect meaning and understanding is therefore rich area of exploration in Arabic-language digital humanities. It is hoped that––as possible––Nuṣūṣ will start to add multiple editions of historic texts in order to facilitate this comparative work.

Image of a table of Arabic-language works held in the Nusus corpus.
Nusus’s “Browse Corpus” page.

Nuṣūṣ’s aspirations lie in providing researchers with an adequate corpus from which to do computational text analysis. To that end, the team has created several different ways for researchers to access and engage with the texts. The “Browse Corpus” feature gives researchers an accurate sense of which specific items are included. If one is looking for a particular author or text, this would be the list to consult. This is also where crucial metadata (information about the item) is located, such as the origin of the digital images (Nuṣūṣ’s own OCR process or the OpenITI project repository), the internal corpus text ID, the date of the historic text’s alleged composition, the discipline, the genre of writing, the title, and the author. Author names link to biographies from the Encyclopaedia of Islam, and titles link to the WorldCat record for the modern edition used in the digitization of the text.

Image of a search for an exact term in the Nusus corpus.
Performing a search for the exact term “عقل” in the Nuṣūṣ corpus.

Furthermore, the Nuṣūṣ team has provided a cross-corpus search tool. Researchers can build a search using the provided fields and Boolean operators (AND, OR), and can specify whether they are searching for an exact term. It is also possible to confine the search to specific titles, authors, or genres. This arrangement encourages researchers to pursue projects that might compare across a scholar’s oeuvre, across a genre of writing (Muslim theology, philosophy, Sufism, or Christian theology), or across a single text. Researchers could use this tool to construct searches across known networks of scholars, as well. As the corpus expands, the ability to conduct searches and collect the resulting data will become increasingly effective and useful.

Readers interested in text and corpora analysis should consult the UT Libraries’ Digital Humanities Tools and Resources guide for more information on methods to apply to corpora like Nuṣūṣ. For recommendations of other corpora that might be useful for your research, consult the Data Set list on the Text Analysis guide. Lastly, as the Nuṣūṣ corpus partners with and derives from the OpenITI repository, it is worth considering the OpenITI repository documentation at the KITAB project. Happy corpus hunting!

Dale J. Correa, PhD, MS/LIS is Middle Eastern Studies Librarian and History Coordinator for the UT Libraries.

[1] The term Islamicate was coined by Marshall G.S. Hodgson in volume 1 of his The Venture of Islam (p. 57).

Read, Hot, and Digitized: KITAB Project Brings Distant Reading to Middle Eastern Studies  

Read, hot & digitized: Librarians and the digital scholarship they love — In this new series, librarians from UTL’s Arts, Humanities and Global Studies Engagement Team briefly present, explore and critique existing examples of digital scholarship.  Our hope is that these monthly reviews will inspire critical reflection of and future creative contributions to the growing fields of digital scholarship.

The KITAB Project, headed by Sarah Bowen Savant of the Aga Khan University, seeks to develop tools and techniques for producing scholarship on text reuse and intellectual networks in the premodern Arabic textual tradition. The project is based on a digital corpus of published texts that represent all genres of writing in Arabic from the earliest works to the beginning of the 20th century CE. Although the corpus draws in part from digital databases of texts, it also relies heavily on digital surrogates of printed volumes which require Optical Character Recognition (OCR) for computational analysis. The KITAB project has partnered with the Open Islamicate Text Initiative to develop an OCR software that has proven more successful than commercially-available products. The collaboration’s published results of this OCR development—called Kraken—can be found here.

A snapshot of initial results using the Kraken OCR software
A snapshot of initial results using the Kraken OCR software

The KITAB project is noteworthy not only for bringing the concepts of text reuse and distant reading to Middle Eastern Studies from a digital humanities perspective, but also for its development of tools designed for Arabic script languages. The needs of right-to-left and non-Roman script languages such as Arabic, Persian, Ottoman Turkish, and Hebrew—namely bidirectionality and non-Roman script recognition capabilities—unfortunately have been neglected to date in key tools utilized by highly successful digital humanities projects. The KITAB project brings the necessity of right-to-left and non-Roman capabilities to the fore by centering the Arabic textual tradition and committing to the development of tools that best meet the needs of the questions asked.

In addition to Dr. Savant, the team behind the KITAB project includes scholars from the U.S. and Europe, notably David Smith (Northeastern University) who developed the passim software upon which the text reuse project is based, and Maxim Romanov (University of Vienna) who heads the Open Islamicate Text Initiative. The team supports the continuing evolution of algorithms that seek to determine which Arabic texts were most quoted, most used by historians, and most commented on over several centuries (roughly 700-1500 CE). These questions might be answered simply enough within one text with a full-text search engine. However, to answer these questions across the Arabic textual tradition requires not only a massive corpus (currently over 4200 items), but also incredible computing power.

The latest KITAB visualization of text reuse across two works attributed to Ibn Qutayba (d. 889 CE).
The latest KITAB visualization of text reuse across two works attributed to Ibn Qutayba (d. 889 CE).

I encourage readers to take a look at the latest text reuse visualization from the corpus, which is based on two works by Ibn Qutayba (d. 889 CE). I also suggest reading Dr. Savant’s critically reflective post on running the passim software across the entirety of the corpus, and the questions raised by the results about intertextuality and what text reuse means in the Arabic context. Lastly, I recommend that those interested and/or involved in the field review information on the KITAB Project’s corpus, including the FAQ links to the Open Islamicate Text Initiative for suggesting new digital titles and new titles requiring OCR. UT Libraries’ collection of historic Arabic texts is one of the largest in the United States and ripe with suggestions for the KITAB corpus (check out this Islamic Empire — History subject heading search to see a sample of UT’s rich Arabic collections).