SYNTACTIC VARIATION IN MASORETIC HEBREW, THE OBJECT CLAUSE RECONSIDERED

In this article we reinvestigate the variation in Masoretic Hebrew of two linguistic features related to the object clause that have diachronic relevance according to various scholars. These features are the order of subject personal pronoun and nominal predicate in certain object clauses, and the variation between יכ and רשא introducing the object clause. We do this by using tools provided by the SHEBANQ project. By examining as many cases as possible throughout the Masoretic Text instead of only a few exemplary cases, we analyse the extent of variation in these syntactic constructions and their relevance for linguistic dating of biblical texts.


INTRODUCTION
In their newest book on historical linguistics and Biblical Hebrew, Rezetko and Young (2014) reject the traditional approach in Hebrew linguistics in which books are grouped a priori on the basis of shared linguistic characteristics. The main point of the book is the philological approach it introduces in which linguistics and textual criticism are considered jointly. In our opinion this is an important methodological step forward. Grouping books in categories such as Early Biblical Hebrew (EBH) and Late Biblical Hebrew (LBH) and comparing other books and texts with these corpora has led to the discovery of many interesting cases of linguistic variation, 1 but this 1 Many examples of features relevant for the study of diachronic variation can be found in Hurvitz (1974Hurvitz ( , 1982, and Hurvitz et al. (2014). A table of features extracted from various important works has been collected in Young, Rezetko and Ehrensvärd (2008), Volume 2, Chapter 4. Diglossia and geographic variation in Biblical Hebrew have been described extensively in the works of Gary Rendsburg (e.g., Rendsburg 1990a, 1990b, and 2002 approach has resulted in various problems as well. On the one hand it is difficult to compare books within each of the subcorpora of EBH and LBH. If books are classified beforehand as LBH, such as Esther and Daniel, it may seem that various linguistic features collected on the basis of the contrast with EBH are characteristic of LBH. However, very often these features occur in only one or two of the core LBH books. 2 Examples of such features are the late lexemes ‫ישט‬ and ‫,תכריך‬ both occurring exclusively in the book of Esther, ‫גזבר‬ (exclusively in the book of Ezra), and ‫תלמיד‬ (exclusively in Chronicles). 3 On the other hand, the traditional approach hides similarities between the subcorpora of EBH and LBH, because the focus lies exclusively on the contrast between these groups of books. In studies working within the framework of linguistic dating the core LBH books are described in terms of what distinguishes LBH from EBH but this is only a relatively thin layer of the language of these books. Of the late lexemes mentioned above, the early alternatives of ‫ישט‬ and ‫תכריך‬ can be found throughout the core LBH books 4 and ‫גזבר‬ and ‫תלמיד‬ do not even have clear early alternatives. 5 This makes that there is hardly any contrast between EBH and LBH concerning these features. 6 An approach that does justice to both continuity and change in Biblical Hebrew (BH) should take into consideration as much data as possible. Electronic editions of the Hebrew Bible and other ancient literature are an important tool for studying the Hebrew Bible and to collect the data. Several packages are available for doing queries in its text. In this article we use the tools of the open source SHEBANQ project. 2 We have described this problem earlier in Rezetko and Naaijer (2016). 3 These late lexemes can all be found in Hurvitz (2014), in which these lexemes are described as being typical of LBH. 4 ‫)יד(‬ ‫שלח‬ is the early alternative of ‫,ישט‬ according to Hurvitz (2014:133). This early alternative is found in the LBH books in Est 2: 21, 3:6, 6:2, 8:7, 9:2, 9:10, 9:15, 9:16, Dan 11:42, Neh 13:21, 1Chron 13:9 and 13:10. 5 Hurvitz (2014:79, 239). The early alternative of ‫תכריך‬ is ‫בגד‬ (Hurvitz 2014: 237), which can be found in LBH in Est 4:1, 4:4, Ezra 9:3, 9:5, Neh 4:17 and 2 Chron 18:9, 18:29, 23:13, 34:19, 34:22, 34:27. 6 More on this issue is described also Rezetko and Naaijer (2016).

Open source tools and the ETCBC database
There is an important requirement that accompanies the use of digital methods in scholarly research: they should not diminish the transparency of the ways that hypotheses are confirmed or rejected. Moreover, computations that lead to new results should be replicable by other scholars. For this reason, it is as essential to publish the data and digital tools as it is important to publish the articles in which the conclusions are stated and discussed. Replication of results obtained by software is not easy in general, because the digital world is in constant flux and all software goes from version to version. To lower the barrier for effective replication we take care that our data are properly archived and our software is available as Open Source in online repositories with versioning. These resources can then be referenced in a persistent way, e.g., through Digital Object Identifiers (DOIs), and they can be freely downloaded.
We base our research on the database of the Eep Talstra Centre for Bible and Computer (ETCBC). A few years ago this database was brought fully online as a research tool in the form of SHEBANQ (System for HEBrew text: Annotations for Queries and markup). On the SHEBANQ website, 7 one can read the complete Masoretic Text according to Biblia Hebraica Stuttgartensia. 8 More importantly, SHEBANQ adds the morphological and syntactic encoding of the ETCBC database in such a way that the user can perform queries on that information.
Much of the information that underlies the present article has been gathered by means of queries in SHEBANQ. With a query one can search for words, phrases, clauses or combinations of these with specific lexical, morphological or syntactic characteristics. These queries have been published and will not change anymore. Even if newer versions of the ETCBC database arrive, it will continue to be possible to view the original list of results in the original version of the data.
Together with the construction of SHEBANQ, two things have been achieved: (i) a research tool has been built to deal with all facets of Hebrew text and data 7 https://shebanq.ancient-data.org. 8 It contains the text of the fourth edition of BHS. In the future we plan to add the text of Hebrew inscriptions, biblical and non-biblical Dead Sea Scrolls. processing: LAF-Fabric; (ii) all relevant resources have been archived and can be downloaded and re-used. We refer to van Peursen et al. (2015) for the data, to Roorda (2015a) for the SHEBANQ software, and to Roorda (2015b) for LAF-Fabric and supporting tools. 9 This makes it possible to reconstruct the whole SHEBANQ later, in case the current website goes out of service. The history of the Hebrew Text database is told in Roorda (2015c) and more about the underlying models for processing text and data can be found in Roorda and van Peursen (2016).

Object clauses in Biblical Hebrew
In the rest of this article we re-evaluate scholarly literature on variation in the structure of the object clause using the tools described above.
The object clause is a complete clause which functions as the object of a transitive verb. Very often this transitive verb is a verb of saying (like ‫נגד‬ or ‫)צוה‬ or a verb of perception (like ‫ידע‬ or ‫.)ראה‬ 10 In this paper we discuss two different linguistic features related to the object clause in MT Hebrew. In section 2 we study the phrase order in the verbless object clause, and in section 3 the variation between ‫כי‬ and ‫אשר‬ introducing the object clause is analysed. In most studies related to linguistic dating of biblical texts the focus is on the late variants occurring in the LBH texts because these are the items that are thought to make it possible to date biblical texts linguistically.
However, in most cases, early variant(s) are found in LBH texts and the late variant(s) are found in EBH texts. This does not necessarily reduce the value of the feature for diachronic investigations, but it makes it necessary to study both the early and late variants throughout the MT, in order to make a balanced evaluation of the distribution of both variants, after which one can get a clearer impression of similarities and differences between EBH and LBH. 9 An overview of all sources is available at https://shebanq.ancient-data.org/sources. 10 Gesenius and Kautzsch (1910: §157), Joüon & Muraoka (2006: §157).

PHRASE ORDER AND RABBINIC HEBREW
In Esther 3.4 one can find the following object clause: The specific order in Esther 3:4 is characteristic of Rabbinic Hebrew and according to some this example can be seen as a forerunner of the Mishnaic use (Bergey 1983:72, Sáenz-Badillos 1993. This is an indication of the diachronic development of Biblical Hebrew (Bergey 1983:72).
To get a more general impression of the this feature, we made a SHEBANQ query that searches for all clauses introduced by ‫כי‬ or ‫,אשר‬ followed by a subject personal pronoun and then a nominal predicate. 12 The list of results contains all the cases of the fixed expression ‫יהוה‬ ‫אני‬ ‫כי‬ and its variations, but besides that it also contains at least two object clauses in the Pentateuch and Former Prophets:

Introduction
A linguistic construction characteristic of Late Biblical Hebrew that has often been cited in the literature on diachrony in Biblical Hebrew is the variation between ‫כי‬ and ‫אשר‬ introducing the object clause. 13 In most cases in the MT ‫כי‬ is used, but sporadically ‫אשר‬ can be found with the same function, e.g., To be able to study the variation between ‫כי‬ and ‫אשר‬ introducing the object clause we created a dataset with the help of LAF-Fabric containing object clauses introduced by ‫כי‬ and ‫.אשר‬ 14

CONCLUSIONS
In this article we discussed the way the object clause has been studied from the perspective of diachronic change in MT Hebrew. For our analyses we have made use of the open source database of the Eep Talstra Centre for Bible and Computer. We have studied three classes of phenomena which have been used to attest a transition from Early to Late Biblical Hebrew, and found the evidence lacking in all three cases.
The first case is the specific order of pronominal subject followed by a nominal predicate in Esther 3:4, which is rare in the Hebrew Bible but common in Mishnaic Hebrew. It is more likely that this is simply an uncommon linguistic feature in the Masoretic Text than that it has diachronic significance, because this word order can be found in object clauses in Early Biblical Hebrew, as we have shown with the help of our SHEBANQ query.
Secondly, a similar case is the supposed diachronic significance of the object clause introduced by ‫אשר‬ in Ezekiel 20:26. The object clause introduced by ‫אשר‬ can be found once in Ezekiel, which is comparable with several EBH books.
Finally we studied the variation between ‫כי‬ and ‫אשר‬ in the core narrative LBH books.
In these books there is a continuity of the use of ‫,כי‬ but there is also a substandard use of object clauses introduced by ‫.אשר‬ These object clauses have a tendency to be governed by other transitive verbs than those that are used generally with object clauses introduced by ‫.כי‬ Although it is difficult to draw strong conclusions based on a relatively small amount of data from an ancient corpus with all its limitations, it is clear that by the use of digital versions in which one can search for morphological and syntactic features, it is possible to make steps forward in the study of the history of Biblical Hebrew. The work by Rezetko and Young (2014) provides a useful theoretical background for further study, of which this article is just one example.