A Quantitative Approach to Variation in Case Inflection in Arabic Documentary Papyri: The Case of 'ab in Construct.

• Author: Kootstra, Fokelien

1. INTRODUCTION

   The use of case is often considered one of the hallmarks of Classical (fasih) Arabic (CAr). Consequently, the lack of case, or a different use of case inflection than the tripartite CAr system, immediately marks a text as non-CAr (e.g., Hopkins 1984: 155). The loss of case is generally put forward as a typical feature of neo-Arabic (e.g., Blau 1977; Ferguson 1959). (1) This loss is often attributed to the sudden need for large numbers of nonnative speakers of Arabic to acquire the language following the conquests (Versteegh 1984; Blau 1977). (2)

   However, recent insights into the history and development of the Arabic language, through the study of pre-Islamic and early Islamic Arabic material, have made it clear that the picture is much more complex. There are attestations of pre-lslamic Arabic that had already largely lost their case system, e.g., Nabataean (Diem 1973), (3) while other forms of Arabic exhibit many options available between those with a complete tripartite case system and those with no morphological case marking at all, e.g., Safaitic, (4) the Quranic Consonantal Text (QCT), (5) and Judaeo-Arabic and certain modern dialects. (6) The Graeco-Arabica from sixth-century Palestine, on the other hand, clearly attest to a loss of case (Al-Jallad 2017: 165-66), while the Psalm fragment, also an Arabic text written in Greek script (Violet 1901) and tentatively dated to the mid-ninth century CE (Al-Jallad 2020: 55-56), only occasionally displays case inflection. (7) This shows that the loss of case is a process that happened in different varieties of Arabic at different times, and that it had already happened in some varieties before the conquests. It also illustrates that this process can have different outcomes.

   Nevertheless, for the Arabic documentary material from the first three centuries AH, a general loss of case inflection is mentioned as one of its characteristic features: "With certain minor exceptions [...] it is quite clear that the language treated in this study was characterized by the absence of a case system, a feature held in common between all varieties of Arabic outside CA[r]" (Hopkins 1984: 155). In light of the variation in case marking in the different Arabic corpora cited above, however, a closer inspection of this very general statement seems warranted. To find out more about the relationship between the loss of case and the language of the papyri I have decided to quantify Simon Hopkins's hypothesis, in order to gain insight into (1) how strongly the language of the documentary papyri is characterized by the absence of case marking, and (2) whether it is possible to find any pattern in the distribution between documents with case marking and those displaying a variety of Arabic with a reduced case system or complete lack of case marking. (8)

   1.1. The Corpus

   In order to test the hypothesis that case was no longer functional in the documentary papyri from the first three centuries AH (622 to 922 CE), the various forms of the noun 'ab in construct attested in papyrus documents dated to this period have been collected from the online Arabic Papyrology Database (APD). (9) The corpus was limited to 'ab in construct, since it is one of five nouns in Arabic that are inflected for case with a long final vowel in construct position. All three cases will be represented orthographically on 'ab in the Arabic papyri, and given the content of the Arabic documents, 'ab is a highly frequent form, providing a sizeable corpus for this preliminary investigation into the distribution of case inflection in the Arabic papyri. Another of the five nouns we can expect to occur with some frequency in the papyri is 'ah. However, including it in this study would have added nowhere near as many examples. (10) In addition to the five nouns, there are other nominal forms that show case inflection, such as indefinite accusative forms, sound plurals, and dual forms; however, none of these distinguishes all three cases and the initial search for these features would have yielded many false positives, so that these forms were also not included in the study. Since case is marked with a long vowel on 'ab in construct, finding evidence for case marking in this environment cannot tell us about the productivity of case marking with final short vowels. It only tells us if and how case functioned in this specific phonological environment.

   In order to collect all forms of 'ab in construct, the textual search function provided by the APD was used, specifically the search option "wildcard characters at beginning," to allow the inclusion of forms with an attached preposition, and "wildcard characters at the end," to allow the inclusion of forms with a suffixed pronoun. With these parameters, I searched for [phrase omitted], and [phrase omitted]. (11) Finally, to include shortened forms, another search was performed for [phrase omitted], [phrase omitted], and [phrase omitted], with the same search options. This added another seven examples to the corpus, (12) bringing the total number of forms to 1,027. The false positives, such as 'abi forms that consisted of noun + first person pronominal suffix, were filtered out by hand. After this, both the form as it was written in the document and its syntactic function, based on the surrounding text, could be determined for 828 of these examples. (13)

   Based on Hopkins's study and his positing that '"ab "father" in status constructus and pronominalis may be inflected as in CA[r] [...] but commoner by far is the unchangeable 'abu in all syntactical positions" (1984: 156), I expected to find an overwhelming amount of the form 'abu (') regardless of its syntactic position.

2. FINDINGS

   Contrary to the claim that uninflected 'abu is by far the most common form, I found that 'ab in construct state is fully inflected for case in the vast majority of the examples (see Table 1 and Fig. 1). Whenever 'abu is used in citation form or subject position, it is impossible to tell whether this is the result of invariable 'abu (which is the common form to get leveled in varieties of Arabic that lose case marking) or a deliberate choice. (14) Therefore 'abu forms are not included in Fig. 1 so as not to obscure the overall view (see Table 1 for an overview of the attestations including 'abu forms). Of the 698 instances of 'ab in construct state in the sample not including 'abu forms, just over eighty percent was fully inflected, following a tripartite case system.

   To keep the labels in figures and tables concise and readable, a distinction will be made between form and function of the various attestations of 'ab. The form refers to their orthographic representation [phrase omitted], and [phrase omitted]. Function refers to a form's morphosyntactic function--nominative, genitive, or accusative--in a tripartite case system. This is not meant to indicate that deviations from a tripartite case system similar to CAr are mistakes, but aims to take the most archaic attested system as starting point and look at developments from there (see Al-Jallad and van Putten 2017 for case as a Proto-Arabic feature). (15)

   The 'abu forms in subject position and in citation form have been included in Table I to show the complete set of data. The numbers relating to 'abu in subject position or citation form have been marked in gray to show that they indicate something slightly different than the other two categories.

   The 19.77 percent of cases, with a different pattern of inflection for 'abu, show that not all the authors of these texts used a written form with a fully inflected case system. Moreover, not all cases seem to have been equally stable, with a markedly lower rate of the use of 'aba in environments typically associated with the accusative in a tripartite case system than of 'abi in typical genitive environments (Table 1). In general, however, we can say that the tripartite case system was the main accepted system in writing. The status of this system as at least somewhat normative seems to be supported by texts in which the authors use case inconsistently even within the same construction, or ones that inflect 'ab in ways that seem like hypercorrections. (16)

   In the following I will begin to untangle the variation attested in this corpus. To this end I first focus on the bulk of the data, using a full tripartite case system, and look at the distribution of this system across several extralinguistic categories, such as time, place, and text type, to give a better idea of what underlies the attested variation in case marking. Once we have an idea of the distribution of the use of case across the corpus, I will zoom in on the forms that are not used according to a tripartite case system and discuss some of the regularities found in their distribution.

3. DISTRIBUTION OF TRIPARTITE INFLECTION

   Since the system distinguishing three different cases is represented in the bulk of the material, it seems worthwhile to see how this system is distributed across several nonlin-guistic features of the corpus. Comparing the use of case in different genres of documents, in different periods, regions, and formulaic contexts, could shed light on the choices behind using one inflectional system or the other, or bring out developments in the written norms behind the Arabic documents written on papyrus.

   3.1. Effect of Genre

   If there is indeed a difference between spoken and written register, as the documents with variable case marking suggest, then we might expect to find a difference in case treatment in different genres of documents. If there was a substantial difference between spoken language and the written norm, we would expect to see more complete case marking in the more formal documents and in documents written by official scribes--those labeled official documents, contracts, land leases, marriage contracts, official letters, orders, receipts for payment or delivery, testimony, and land registration...