Navigation – Plan du site

AccueilNuméros26Navigating Noise: A Stratified Mo...

Navigating Noise: A Stratified Model for Scholarly Digital Editions of Arabic Manuscripts in Hebrew Script

Valentina B. Lanza

Résumés

Cet article examine le rôle des éditions numériques savantes dans le développement de la recherche textuelle, en se focalisant sur le judéo-arabe, une variété linguistique unique utilisant l’alphabet hébreu pour représenter l’arabe. Nous proposons un modèle stratifié pour les éditions numériques savantes, visant à gérer les complexités orthographiques du judéo-arabe, notamment la transcription, la conversion de l’écriture, la normalisation et l’intégration des métadonnées. L’étude souligne l’importance cruciale de la gestion du « bruit » dans l’édition des textes, particulièrement lors de l’acquisition des données, où toute distorsion ou omission d’éléments orthographiques peut compromettre l’intégrité scientifique. Une étude de cas portant sur un manuscrit médiéval en judéo-arabe illustre l’application de ce modèle, en montrant comment il préserve l’intégrité textuelle tout en enrichissant l’analyse scientifique. Cette méthodologie aborde non seulement les défis spécifiques au judéo-arabe, mais elle est également adaptable à d’autres systèmes allographiques.

Haut de page

Texte intégral

1. Preliminary Remarks

1Scholarly Digital Editions (SDEs) mark a significant advancement in the humanities and social sciences, seamlessly blending digital tools with academic methods to craft, showcase, and delve into textual content. They signify a profound paradigm shift in scholarly practices, opening up unprecedented avenues for exploration, examination, and sharing. As digital technologies continue to evolve exponentially, scholars from diverse disciplines are harnessing the potential of digital platforms to revolutionize traditional approaches to editing and publishing. Yet, this technological wave brings with it a complexity in defining scholarly editing. What is certain is that SDEs are characterized by their commitment to interdisciplinary scholarship, incorporating diverse approaches, methodologies, and materials. This reflects the evolving nature of academia, where disciplinary boundaries are increasingly blurred, fostering collaboration across fields.

2Grounded in the principles of critical editing, SDEs uphold scholarly integrity and rigor: while digital technologies have transformed the landscape of scholarly editing, they strike a balance between innovation and tradition, leveraging digital tools while maintaining core principles of critical scholarship. The term “edition” itself, while commonly associated with published works, takes on a distinctive meaning within the context of scholarly endeavors. It transcends mere publication to encompass a multifaceted process guided by specific theoretical frameworks and methodological principles:

No matter how malleable, diverse and dynamic an edition is, it must be original or, in other words, must add new knowledge [emphasis our]. Work that does not produce new knowledge is considered to be a mere reproduction of the primary source. (Driscoll & Pierazzo 2016: 162)

3The notion of richness or added value, relating to the depth and complexity inherent in SDEs, has been a topic of discussion within the digital editing community. In Robinson (2002: 60-62), it is underscored that the objective of SDEs should not simply entail the proliferation of images and texts within a “hypermedia archive”. Indeed, while digital technologies have facilitated the presentation of information in visually appealing and accessible formats, the genuine richness of critical digital editions lies in their capacity to effectively convey intricate concepts and foster deeper scholarly engagement. Therefore, the concept of richness encompasses both the breadth of information presented and the depth of comprehension it promotes, ultimately enhancing the reading experience and encouraging critical inquiry.

4In the realm of Natural Language Processing (NLP), the notion of richness or added value is closely intertwined with that of noise, representing complementary facets of the same phenomenon. Indeed, within NLP, noise may encompass elements that necessitate removal or normalization to prevent text distortion and maintain system integrity. Simultaneously, it includes elements that merit preservation due to their potential to enrich content, offer supplementary information, or even enhance system performance. In Al Sharou et al. (2021), these two types of noise are classified as “harmful noise” and “useful noise”, respectively. Within the taxonomy of noise outlined in the same work, orthography plays a significant role, particularly in distinguishing between unwanted constructs in text and intentional deviations from language conventions, underscoring the importance of training models to recognize and address orthographic variations effectively in NLP tasks. Orthographic noise may encompass various sub-types, including spelling errors and orthographic variants. Orthographic variants may involve the presence of diacritical marks that, in addition to distinguishing some letters from each other, may alter the meaning of words, as seen in Arabic (Al Sharou et al. 2021: 55-56).

  • 1 Numerous scholarly works delve into this subject, encompassing diverse linguistic contexts and view (...)

5In the context of digital language processing and text representation, attention to orthography emerges as a crucial element for ensuring accurate and high-quality results.1 This concept is particularly relevant in cases of allography, where a language uses an alphabet that typically corresponds to another language through its own system of correspondence. In such situations, correct orthography management becomes even more essential, as the specific orthographic conventions of both languages involved must be considered to ensure an accurate and understandable representation of the digital text.

6Within the general framework of the allography, in our study we will focus on Judeo-Arabic (JA), which is the allographic language variety comprising Arabic written in the Hebrew abjad.

7Aiming at filling a notable gap in the scholarly literature regarding this textual tradition, we propose a model for constructing SDEs tailored to the linguistic complexities of this unique linguistic variety, specifically addressing its orthographic specificity. This scenario, which amalgamates the tools of digital philology and their potential integration for the treatment of JA, has garnered limited attention in academic discourse, despite the significant position that JA occupies within the spectrum of Jewish languages due to its historical depth, cultural significance, and linguistic richness (see below).

8This article pursues two main research goals: 1) to offer a methodological reflection and possible solutions to adopt when working on JA manuscripts and editions, and 2) to present our digital edition model, which takes into account all the stratification of problems that characterize JA texts. An application will also be presented as a case study. We emphasize that although the proposed model is tailored for JA manuscripts, it can readily be adapted for various forms of allography. Our examination encompasses three categories of noise delineated within the scope of this volume: noise during data collection and recording, data preparation and pre-processing, and the annotation process and metadata. Our focus lies on the first category, in fact, emphasizing the significance of orthographic preservation when handling JA texts, we argue that any distortion or omission of orthographic elements during the data acquisition stage constitutes noise, which undermines the integrity of scholarly endeavor.

2. Object of Study: JA as a Unique Example of Allography

  • 2 For a thorough understanding of JA, including its historical periods of written usage, linguistic f (...)

9JA is a linguistic phenomenon resulting from the cultural intersection between Jewish and Arab-Islamic civilizations, particularly during the period of significant Arab-Islamic conquests starting in the 7th century CE. JA emerged as Jews residing in newly conquered territories began adopting Arabic, integrating it into their linguistic framework while also conforming it to their own language system. Over time, JA developed into a distinct linguistic variety characterized by its unique features and adaptations. Key characteristics of JA include its use by Jews for a Jewish readership and its representation in Hebrew script. However, it is essential to recognize that JA is not a homogeneous entity but rather exhibits polyhedral complexity, necessitating a nuanced understanding. Such a complexity arises from various factors, including a) its temporal extension and the vast corresponding literature, and b) its mixed linguistic features. Below, we briefly explore these factors that contribute to JA’s multifaceted and diverse linguistic nature.2

  • 3 For JA periodization, scholars typically distinguish between Medieval and Post-Medieval (or Classic (...)
  • 4 Goitein explores the idea of “Jewish-Arabic creative symbiosis”, focusing on the period from 500 CE (...)
  • 5 On the history of JA literature, see Tobi (2011).

10The temporal span of JA was particularly extensive, encompassing a period of fourteen centuries if we consider its more contemporary and folkloristic uses.3 Its wide geographical spread and the substantial body of significant literature over the centuries contribute to its unique position within the spectrum of Jewish languages –which includes Yiddish, Judeo-Persian, Judeo-Greek, Ladino, among others. Jewish scholars and intellectuals in the medieval Islamic world produced a vast body of literature in JA, which encompasses a wide range of genres including translations and commentaries on the Hebrew Bible, Talmudic texts, works of Jewish philosophy and mysticism, legal writings, poetry, and secular prose. This literature played a crucial role in shaping the intellectual landscape of both Jewish and Arabic-speaking societies, influencing scholars and thinkers across religious and cultural boundaries by blending elements of Jewish religious thought with Arabic literary forms, thus creating a distinctive fusion of languages and cultural perspectives.4 A significant portion of JA literary works was brought to light through the discovery of the Cairo Genizah in the late 19th century.5 The thousands of documents and fragments unearthed from this repository constitute a paramount source for understanding the social, political, economic, and religious milieu of ancient Cairo in particular between the 10th and the 13th centuries. Moreover, they offer invaluable insights into the dynamics of coexistence among the three major monotheistic faiths –Christianity, Islam, and Judaism– present in Egypt during that era.

11Linguistically, JA shares certain traits with Neoarabic dialects, such as a tendency towards analytical constructions and modifications in the inflectional system. Nonetheless, it maintains its distinctiveness through the incorporation of Hebrew and Aramaic lexical and grammatical elements, as well as its predominant use of Hebrew characters in writing, including specific orthographic conventions such as word-final letter forms and occasional vowel markings. In terms of its written representation, JA relies on a correspondence system between the Hebrew and Arabic abjads. Since the Hebrew abjad includes twenty-two consonants versus the twenty-eight of the Arabic, JA authors used to mark consonants with diacritical marks. Nevertheless, this system has never been truly standardized, and JA texts exhibit a high degree of variability in terms of both letter correspondence and the use of diacritical marks across the various periods.

12The analysis of orthography in Arabic texts written in Hebrew script is particularly important for investigating socio-cultural implications as well as linguistic phenomena. Indeed, if we consider language as a general expression of identity, the use of Hebrew characters represents the partial detachment of JA culture from the surrounding environment (Blau 1999: 38). However, historically, the use of Arabic by Jews, even when written in their own abjad, has been indicative of the level of absorption of Jewish communities into Arab-Islamic society. Indeed, due to increased isolation of the Jews and reduced contact with Muslim language and culture, JA underwent significant changes during the 15th century. Orthography and the entire linguistic system shifted towards a more “Hebrew-like” variety and authors increasingly chose Hebrew over Arabic for their works.

13Considering linguistic implications, it is notable that JA orthography had to adapt to the absence of specific Arabic phonemes in the Hebrew abjad and often relied on the use of scriptio plena. Consequently, a comprehensive analysis of the orthography reveals intriguing linguistic phenomena (Hary 1990: 77). In fact, studying allographic texts offers distinct advantages, particularly in uncovering phonological phenomena, which are crucial for understanding the language’s evolution. In our case, this sheds light on the development of Arabic.

3. Methodology: The Stratified Model

14Our model is based on the essential notion that preserving the orthographic integrity of JA texts is paramount. Transcription and matters concerning orthographic treatment are fundamental aspects of textual criticism. Proceeding from the premise that “transcription itself is a representation”, and that “representation is a necessity for an edition. Critical engagement without representation is not an edition –but an examination, a catalogue or a description” (Driscoll & Pierazzo 2016: 24), it becomes apparent that the establishment and implementation of guidelines in this respect constitute a pivotal process.

  • 6 See, e.g., Hopkins & Blau (1984), Hary (1990) and Lanza (2020).

15As previously mentioned, orthography in JA serves as a foundational element and underpins any subsequent analysis, be it linguistic, historical, literary, or otherwise. However, the inherent orthographic specificity of JA presents a notable accessibility challenge: accessing JA texts requires proficiency not only in Arabic but also in Hebrew, and merely possessing basic reading skills in Hebrew may not suffice, given the occurrence of Hebrew and Aramaic lexical and grammatical elements within the texts. Therefore, texts written in JA are not directly accessible/readable, even for Arabists, and this greatly limits their investigation and dissemination. Such a complexity may elucidate why editions of JA texts are frequently presented directly in Arabic script, sometimes without adequately detailing the correspondence system used for alphabetic transduction. This inclination toward normalization within philological work likely also arises from ideological factors. Over time, in fact, the academic studies have often treated JA as a language to be corrected, a hybrid form requiring standardization. Consequently, insufficient attention has been given to the essential textual data. This ideological bias could explain the absence of proper systematization in philology concerning scholarly editions of JA texts, despite numerous linguistic studies highlighting the importance of orthography in JA.6

16To mitigate the noise presented by the common practice of presenting JA texts in a “normalized” form, we suggest implementing a stratified model for handling JA manuscripts.

Figure 1. Standard model for editions of Judeo-Arabic texts

Figure 1. Standard model for editions of Judeo-Arabic texts

17The model is depicted in Fig. 1, and it includes one mandatory Level (Level 1, highlighted in red) and three optional ones. As can be observed, Levels 1-3 primarily address orthographic concerns, whereas the information contained in Level 4 tends to be varied and closely aligned to the research interests of the specific project. Let’s delve into this stratified model in some detail.

  • Level 1. This pertains to data acquisition and entails accurately transcribing the manuscript text, preserving each word and letter faithfully.

  • Level 2. After transcribing the text in its original form, it is permissible to convert it from Hebrew to Arabic script. However, it is important to note that this conversion involves a straightforward transposition of letters, maintaining a 1:1 correspondence. Since the correspondence system between the Hebrew and Arabic abjads is not univocal in JA, this step can be approached in two ways:
    a) utilizing a correspondence system inferred from the text itself;
    b) employing another acknowledged correspondence system in JA.
    Regardless of the approach chosen, it is imperative to clearly delineate the system used.

  • Level 3. The third Level allows us to virtually normalize the Arabic, including vowels and diacritical marks.

  • Level 4. The fourth Level corresponds to metadata integration. This may encompass additional information such as phonetic transcription, translation, specific linguistic details on textual portions, external links, etc. In this perspective, it is potentially limitless.

18If Level 2 is propaedeutic to Level 3, dealing with the transduction of Hebrew characters into Arabic characters, and eventual normalization, Level 4 can be considered self-consistent instead. Indeed, an edition featuring only Levels 1 and 4, a faithful transcription of the text alongside supplementary data, is still valid.

19We also stress that, while Levels 1, 2 and 3 are categorized sequentially as well as by relevance, what we referred to as Level 4 is not strictly sequential compared to the previous ones, in that metadata can be added during the completion of each Level. In this context, our methodology involves the execution of in-line mark-up across various stages of work without compromising its effectiveness, while also managing potential metadata superimposition.

20Using this model, no data is lost and no “harmful noise” is generated: everything subsequently included becomes an added value. In fact, it considers the multifaceted complexity of JA linguistic variety and allows for the potential representation of all its nuances and additions that could enhance its study and comprehension.

4. Case Study

21In this section, we describe a SDE of a JA text found in a 15th-century manuscript: The Jewish Theological Seminary, NY, Ms.5491. The manuscript consists of an incomplete book of the Writings. The text we focused on is trilingual, comprising the Song of Songs, its Aramaic Targum, and a JA translation of the Song, accompanied by a commentary in JA. While the trilingual text is thoroughly examined for literary contextualization, the edition primarily focuses on the JA portion.

  • 7 www.jadi-project.com.
  • 8 Further developments of this edition, towards the inclusion of additional Levels according to our m (...)

22This SDE has been developed within the JADI project,7 funded by Sapienza University of Rome. The primary goal of JADI project is to promote the study and dissemination of JA documentation, facilitated by the use of new technologies for the creation of digital editions. Aligned with the objectives delineated within the JADI deliverables, the process of transcribing the text into Arabic characters was deemed extraneous to the research interests, which primarily pertained to the linguistic analysis of the text, focusing, in particular, on grammatical and lexical elements. Consequently, this edition well represents an example of the simplified version of the stratified model, addressing Levels 1 and 4 in the sense previously described.8

  • 9 TEI is the de facto standard for digital textual representation, enabling standardization of survey (...)
  • 10 Today, TEI is well-suited for editing Semitic texts, addressing challenges like complex morphology (...)

23The starting point, therefore, was the faithful transcription of the text as it appeared in the manuscript. During this phase, particular attention was paid to specific elements that, if overlooked, would introduce noise, such as typos, repetitions, annotations, deletions, and additions. Preserving all these details is crucial for maintaining the integrity of the digital edition. Indeed, this information can be regarded as “useful noise” because it can contribute to enriching the edition at various levels of information. The transcription was performed using XML to adhere to the Text Encoding Initiative (TEI) standards,9 allowing for the addition of metadata, specifically to fully implement Level 4 according to our methodology.10

  • 11 The sub-tags for identifying languages have been elaborated by the Internet Assigned Numbers Author (...)
  • 12 We specify that the <milestone> tag was used, not strictly conventionally in this case, since Songs and verses</milestone> (...)

24The initial phase of the encoding process focused on delineating the text structure and internal subdivisions (material divisions). This encompassed the encoding of page (recto and verso) and line numbers, which are crucial for textual description and the organization of the digital edition. Although the JA portion of the Song of Songs is the primary focus of this study and edition, at an early stage the text was fully transcribed. We used the <seg> element to encode the three different languages. These, were then specified using the @xml:lang attribute and the codes ‘he’, ‘jpa’, and ‘jrb’, representing Hebrew, Aramaic (Jewish Palestinian Aramaic), and JA, respectively.11 This encoding enabled us to extract only the JA section while maintaining a complete transcription of the entire text. Alongside the trilingual portions of the text, other interpretive divisions have been encoded, including the Songs and verses using the <milestone> elements. Fig. 2 provides an extract of the encoding, illustrating both material and interpretive internal subdivisions.12

25For the purpose of our linguistic analysis, we focused on two elements: the occurrence of non-Arabic terms, and the quotations from other biblical books or other texts from the Judaic tradition. As already mentioned, the occurrence of non-Arabic terms involves Hebrew and Aramaic elements. Given that the Hebrew within JA texts largely stems from the post-Biblical era, it naturally integrates considerable Aramaic influences. Therefore, for our study, we rely on the insights provided by Joshua Blau and we regard Hebrew and Aramaic as an “indivisible unit” (Blau 1999: 133). This includes an implicit reference to Aramaic whenever we address the occurrence of Hebrew elements.

26Non-Arabic terms are encapsulated within the <foreign> element (see Fig. 3), while biblical quotations are enveloped within the <quote> element (see Fig. 4). For non-Arabic terms, a reference to an external dictionary is provided using the @corresp attribute within the <foreign> element, with the value pointing to a specific entry in an external XML file (dict.xml). Similarly, biblical quotations are annotated with the @corresp attribute within the <quote> element, linking to a particular location in another external XML file (bibl.xml).

Figure 2. Example of the encoding of material and interpretive internal subdivisions

Figure 2. Example of the encoding of material and interpretive internal subdivisions

27Utilizing both the <foreign> (and <quote>) element and the @corresp attribute yields multiple benefits in scholarly encoding. The <foreign> (and <quote>) element aids in content identification and semantic representation, thereby improving the readability and interpretability of the encoded text. Meanwhile, the @corresp attribute ensures precise referencing, facilitating cross-referencing and retrieval of relevant information, enabling exact linking to external resources like dictionaries or other texts. This combination of explicit markup and precise linking enhances the interoperability of the encoded text and allows for the potential integration of additional information and metadata from external files into the digital edition. This integration guarantees that the digital edition remains comprehensive and enriched with supplementary data.

Figure 3. Example of the encoding of non-Arabic terms

Figure 3. Example of the encoding of non-Arabic terms

Figure 4. Example of the encoding of quotations

Figure 4. Example of the encoding of quotations
  • 13 EVT originated within the framework of the Digital Vercelli Book project (http://vbd.humnet.unipi.i (...)

28In terms of visualization, our choice fell upon the Edition Visualization Technology (EVT), an open-source tool developed by Roberto Rosselli del Turco and his team specifically for constructing digital editions from TEI XML-encoded texts.13 The version of EVT utilized is the latest release, EVT2 beta, specially adapted and extended for the specific needs of the JADI project.

29This last step formally concludes the SDE of the considered JA manuscript. Employing faithful transcription and XML encoding that adhere strictly to TEI standards, we uphold the manuscript’s integrity while augmenting it with metadata to deepen comprehension.

5. Concluding Remarks

30In this article, we have introduced a stratified model designed to manage Arabic manuscripts written in Hebrew script, thereby addressing a significant gap in philological studies, especially within digital philology. Our model tackles the intricate linguistic challenges presented by JA texts, a scholarly domain historically underexplored despite its profound cultural, historical, and linguistic significance. By focusing specifically on orthographic nuances, the model aims to preserve the authenticity of original texts while utilizing advanced digital philology tools for meticulous analysis and presentation of manuscripts. This interdisciplinary approach promotes collaboration across diverse academic fields, including linguistics and digital humanities, and enhances understanding of the cultural and linguistic dynamics inherent in Arabic texts written in Hebrew script. Central to our model is the initial phase of faithful transcription of the manuscript, meticulously retaining every spelling detail, even if it is “non-standard”. Only after this, the focus can be put on the normalization of spelling, eventually integrated with metadata and varies according to research interests. We addressed the issue of noise according to the specifics of our methodology, describing “harmful noise” as an omission rather than an addition of information: the omission of original elements found in the manuscript, often resulting from the direct normalization of the text. In contrast, we argued that a faithful transcription, as the initial step of any edition, is essential. This approach ensures the preservation of all elements that enrich the textual content, provide context, or enhance understanding, constituting what we term as “useful noise” as they improve the edition at various levels of information. Through practical application in a case study, we showed how creating accessible and comprehensive digital editions. This scheme offers a clear and methodological structure adaptable to address the specific challenges of transcribing and representing allographic texts in general, characterized by distinct orthographic features.

Haut de page

Bibliographie

Ahmed M. A. H. (2018). “XML Annotation of Hebrew Elements in Judeo-Arabic Texts”, Journal of Jewish Languages 6(2): 221-242. https://doi.org/10.1163/22134638-06021122.

Al Sharou K., Li Z. & Specia L. (2021). “Towards a Better Understanding of Noise in Natural Language Processing”, in R. Mitkov & G. Angelova (eds.) Proceedings of the International Conference on Recent Advances in Natural Language Processing (RANLP 2021), 53-62. INCOMA Ltd. https://aclanthology.org/2021.ranlp-1.7.

Besdouri F. Z., Mekki A., Zribi I. & Ellouze M. (2021). “Improvement of the COTA-Orthography system through language modeling”, 2021 IEEE/ACS 18th International Conference on Computer Systems and Applications (AICCSA), 1-7. https://doi.org/10.1109/AICCSA53542.2021.9686898.

Blau J. (1999). The Emergence and Linguistic Background of Judaeo-Arabic: A Study of the Origins of Middle Arabic (3rd ed.). Jerusalem: Ben-Zvi Institute.

Chakravarthi B. R., Rani P., Arcan M. & McCrae J. P. (2021). “A Survey of Orthographic Information in Machine Translation”, SN Computer Science 2(4): 330. https://doi.org/10.1007/s42979-021-00723-4.

Clivaz C., Schulthess S. & Sankar M. (2017). “Editing New Testament Arabic Manuscripts in a TEI-base: Fostering close reading in Digital Humanities”, Journal of Data Mining & Digital Humanities, Special Issue on Computer-Aided Processing of Intertextuality in Ancient Languages. https://doi.org/10.46298/jdmdh.1390.

Driscoll M. J. & Pierazzo E. (eds.). (2016). Digital scholarly editing: Theories and practices. Cambridge, UK: Open Book Publishers.

Eskander R., Habash N., Rambow O. & Tomeh N. (2013). “Processing spontaneous orthography”, Proceedings of the 2013 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, 585-595.

Goitein S. D. (1955). Jews and Arabs. Their Contacts Through the Ages. New York: Schocken Books.

Hary B. (1990). “The importance of the orthography in Judeo-Arabic texts”, in Proceedings of the Tenth World Congress of Jewish Studies. Jerusalem: World Union of Jewish Studies, 77-84.

Hary B. (1992). Multiglossia in Judeo-Arabic. With an edition, translation and grammatical study of the Cairene Purim scroll. Leiden: Brill.

Hopkins S. & Blau J. (1984). “On Early Judaeo-Arabic Orthography”, Zeitschrift Für Arabische Linguistik 12: 9-27.

Khan G. (2016). “Judeo-Arabic”, in L. Kahn & A. D. Rubin (eds.) Handbook of Jewish languages. Leiden: Brill, 22-63.

Lancioni G. & Joosse N. P. (2016). “The Arabic Diatessaron Project: Digitalizing, Encoding, Lemmatization”, Journal of Religion, Media and Digital Culture 5(1): 205-277. https://doi.org/10.1163/21659214-90000075.

Lanza V. B. (2020). “Judeo-Arabic Orthographies: Insights from a Fifteenth-Century Šarḥ”, Eurasian Studies 18(1): 134-148.

Olivieri S. (2023a). The Kitāb al-ʾInṣāf by Ibn al-Anbārī in electronic form [Dataset]. Freie Universität Berlin. https://doi.org/10.17169/REFUBIUM-41299.

Olivieri S. (2023b). The Kitāb Sībawayhi in electronic form [Dataset]. Freie Universität Berlin. https://doi.org/10.17169/REFUBIUM-41300.

Robinson P. M. W. (2002). “What is a critical digital edition?”, Variants: The Journal of the European Society for Textual Scholarship 1: 43-62. https://doi.org/10.5281/ZENODO.6533168.

Saadane H. & Habash N. (2015). “A Conventional Orthography for Algerian Arabic”, in N. Habash, S. Vogel & K. Darwish (eds.) Proceedings of the Second Workshop on Arabic Natural Language Processing. Association for Computational Linguistics, 69-79. https://doi.org/10.18653/v1/W15-3208.

Stillman N. A. (2010). “Judeo-Arabic—History and Linguistic Description”, in N. A. Stillman (ed.) Encyclopedia of Jews in the Islamic World. Consulted online on 6 February 2024, http://dx.doi.org/10.1163/1878-9781_ejiw_COM_0012320.

TEI Consortium (ed.). (2023). TEI P5: Guidelines for Electronic Text Encoding and Interchange. (Version 4.7.0). TEI Consortium.

Tobi Y. (2011). “Literature, Judeo-Arabic”, in Encyclopedia of Jews in the Islamic World (Vol. 3, 271-278). Leiden: Brill.

Zribi I., Boujelbane R., Masmoudi A., Ellouze M., Belguith L. H. & Habash N. (2014). “A Conventional Orthography for Tunisian Arabic”, Proceedings of the Ninth International Conference on Language Resources and Evaluation (LREC-2014), Reykjavik, Iceland, May 26-31, 2014, 2355-2361. http://www.lrec-conf.org/proceedings/lrec2014/summaries/219.html.

Haut de page

Notes

1 Numerous scholarly works delve into this subject, encompassing diverse linguistic contexts and viewpoints. This comprehensive exploration encompasses various domains within artificial intelligence, focusing on linguistic analysis and implementations of computer technologies for text and digital material management, publication, and analysis (on machine learning see, e.g., Chakravarthi et al. 2021). On Arabic orthography treatment see, e.g., Eskander et al. 2013, Zribi et al. 2014, Saadane & Habash 2015, Besdouri et al. 2021.

2 For a thorough understanding of JA, including its historical periods of written usage, linguistic features, and an analysis of its spoken dialects, refer to Khan’s comprehensive work (Khan 2016).

3 For JA periodization, scholars typically distinguish between Medieval and Post-Medieval (or Classical and Modern) periods, before and after the 15th century, respectively, following the Jewish expulsion from Spain (Stillman 2010). A more detailed periodization, as identified by Hary, includes five phases: Pre-Islamic JA, Early JA, Classical JA, Later JA, and Contemporary JA (Hary 1992: 75).

4 Goitein explores the idea of “Jewish-Arabic creative symbiosis”, focusing on the period from 500 CE to 1300 CE. This era marks the beginning and early development of Islam within its Jewish context, examining how Islam influenced Jewish intellectual life and how Arabic language and literature impacted Hebrew tradition (Goitein 1955: 11).

5 On the history of JA literature, see Tobi (2011).

6 See, e.g., Hopkins & Blau (1984), Hary (1990) and Lanza (2020).

7 www.jadi-project.com.

8 Further developments of this edition, towards the inclusion of additional Levels according to our methodology, is in progress.

9 TEI is the de facto standard for digital textual representation, enabling standardization of survey methods and resource reusability. It is important to note that in TEI, elements are encapsulated by angle brackets <>, and each must be properly terminated with a closing tag </>. Attributes are designated with @ and are associated with specific element classes. Milestones, represented by </>, consist of unique identifiers inserted directly into the text without the need for both opening and closing tags, contrasting with the structured nature of “element” classes. Find additional information about the TEI consortium on https://tei-c.org/.

10 Today, TEI is well-suited for editing Semitic texts, addressing challenges like complex morphology and right-to-left scripts. TEI guidelines incorporate tools for precise linguistic annotation, manuscript description, and non-Latin script management. Numerous projects demonstrate TEI’s efficacy in editing Semitic texts, offering frameworks and resources for digital research in Semitic languages. To cite a few examples within the domains of Hebrew, Arabic, and Judeo-Arabic consider, for instance, Olivieri (2023a, 2023b); Ahmed (2018); Clivaz et al. (2017); Lancioni & Joosse (2016).

11 The sub-tags for identifying languages have been elaborated by the Internet Assigned Numbers Authority (IANA). The authoritative sub-tags list as finalised by IANA is available at http://www.iana.org/assignments/language-subtag-registry. See TEI Consortium (2023: xlvi).

12 We specify that the <milestone> tag was used, not strictly conventionally in this case, since Songs and verses are considered fundamental elements within the text, crucial for navigation and interpretation.

13 EVT originated within the framework of the Digital Vercelli Book project (http://vbd.humnet.unipi.it/beta2/). However, it has since developed into a versatile tool capable of accommodating various texts and requirements. For instance, it is currently being employed for HumaRec project (https://humarec.org/), which involves the publication of a unique trilingual (Greek-Latin-Arabic) New Testament manuscript. More info on EVT is available at http://evt.labcd.unipi.it/.

Haut de page

Table des illustrations

Titre Figure 1. Standard model for editions of Judeo-Arabic texts
URL http://journals.openedition.org/corpus/docannexe/image/8839/img-1.png
Fichier image/png, 86k
Titre Figure 2. Example of the encoding of material and interpretive internal subdivisions
URL http://journals.openedition.org/corpus/docannexe/image/8839/img-2.png
Fichier image/png, 80k
Titre Figure 3. Example of the encoding of non-Arabic terms
URL http://journals.openedition.org/corpus/docannexe/image/8839/img-3.png
Fichier image/png, 362k
Titre Figure 4. Example of the encoding of quotations
URL http://journals.openedition.org/corpus/docannexe/image/8839/img-4.png
Fichier image/png, 328k
Haut de page

Pour citer cet article

Référence électronique

Valentina B. Lanza, « Navigating Noise: A Stratified Model for Scholarly Digital Editions of Arabic Manuscripts in Hebrew Script »Corpus [En ligne], 26 | 2025, mis en ligne le 14 janvier 2025, consulté le 08 février 2025. URL : http://journals.openedition.org/corpus/8839 ; DOI : https://doi.org/10.4000/1364p

Haut de page

Auteur

Valentina B. Lanza

University of Naples “L’Orientale”

Haut de page

Droits d’auteur

Le texte et les autres éléments (illustrations, fichiers annexes importés), sont « Tous droits réservés », sauf mention contraire.

Haut de page
Rechercher dans OpenEdition Search

Vous allez être redirigé vers OpenEdition Search