The Textual History of the Ethiopic Old Testament Project (THEOT): Goals and Initial Findings 1

This article offers an introduction to the Textual History of the Ethiopic Old Testament (THEOT) project.This includes a description of the background to THEOT and its primary purpose of mapping the history of the transmission of the Ethiopic Old Testament. The bulk of the article summarizes the project’s preliminary findings, generally, and, in particular, about Ethiopic Psalms, Song of Songs, Deuteronomy, Ruth, Amos,

assefa et al.
Textus 29 (2020) 80-110 as reliable presentations of that text. Preliminary data demonstrate that from the fifteenth to the twentieth century there are at least three major stages of development in the Ethiopic Old Testament tradition. It seems prudent to suppose that the previous millennium would have seen at least a similar number of major developments in its history. We have the advantage of hindsight, though, and do not judge previous generations of scholarship for what they could not have known.

Goals
It was the microfilm projects of the 1970s, 1980s, and 1990s, that began to change the possibilities for textual studies of the Ethiopic Bible, and since the turn of the twenty-first century, the dramatic increase in manuscript digitization has brought us squarely into a new position. By 2011, we could with relative ease access and study more than thirty manuscripts of every book of the Ethiopic Old Testament. It was at this time that the THEOT project began the work of transcribing manuscripts of Ethiopic Obadiah, our pilot project. Six years later, we have substantial sets of manuscript transcriptions (meaning 30+) for twenty-four books of the Ethiopic Old Testament and are well on the way to gaining at least some general ideas about the nature of the tradition. The goal of the THEOT project is to transcribe and statistically analyze variations within thirty or more manuscripts of each book of the Ethiopic Old Testament in order to describe the textual history of each book and to characterize each of its major phases. It is important to note that our goal is not to reconstruct the earliest attainable text, but to understand Ethiopia's history of biblical transmission from the earliest attested text up to today. It is also not part of our goal to produce critical editions of each book, but to lay the foundation for others to do such work.4

1.1
The Primary Text Historical Purpose of the Project Textual traditions offer at least two objects of study: the origins of the tradition(s) and the emerged form(s) of the tradition(s). The search for a tradition's origins is an indispensable task whether pursuing the form of the initial text, first translation, or the Vorlage behind that first translation.5 Likewise, the study Textus 29 (2020) 80-110 of the history of the text illuminates the ever-emerging forms that the text underwent. In particular, those forms that came to dominate certain epochs and regions demand special notice. In theory, we might think it impossible to identify any points in the copying history of a text that are qualitatively different from any other moment of copying. But there are clearly moments in each textual history where forms of the text emerged that gained an ascendancy over the tradition for a limited period. In small and larger ways, the text would continue to change, until another flux moment occurred, when the dominating form of the tradition was replaced by a new majority form of the text. It is precisely on this aspect of the tradition that our project focuses. First, it seeks to identify the forms of the text that dominated certain epochs. Second, it wants to reconstruct and understand the nature of these processes of emergence. What do they tell us about the perceived inadequacy of a previous text form? What types of changes emerged in the processes of transmission history and why?
The study of variants is carried out slightly differently when reconstructing the initial text and when studying subsequent text forms. When reconstructing the initial text, the text critic must produce a stemma, which in part requires making a judgment about the text-critical value of each variant. To put forward a stemma is to identify the extant variants and propose a theory explaining how the variants relate to one another, specifically to describe how one gave rise to the others. Variants that could arise independently through the same common scribal error (polygenesis) do not contribute much to this process, for they cannot be used to prove a genetic relationship among manuscripts.
For the text-historical enterprise, the value of variant readings is measured differently. Although it is helpful to identify the possible forces that gave rise to a text form, in the end the emerged forms of the text are just as important as the generative forces that shaped them. In fact, what is important is not how they originated, but how they gained ascendancy for a certain period. Toward that end, all shared variants are significant text-historically, even if not textcritically. If certain readings characterize a certain text form, regardless of how they arose, they will never be discarded as insignificant for textual history; they are precisely of interest for textual history.
talists-the primary group of Western scholars that have invested much time in the study of the Ethiopic Old Testament-have been interested in the Greek form behind that version. See, for instance, Bachmann's Greek retroversion of the Ethiopic found in Die Klagelieder Jeremiae in der aethiopischen Bibelübersetzung: Auf Grund handschriftlicher Quellen mit textkritischen Anmerkungen (Halle: Max Niemeyer, 1893). assefa et al.
Textus 29 (2020)  We mention this last point with a view to the final portion of this article in which we set forth characteristic readings of certain clusters of manuscripts in the textual history of a particular book of the Ethiopic Old Testament. Even though we occasionally comment on scribal error as the primary cause for variant readings, these comments are not intended to be bases for setting these readings aside as text-historically insignificant.

1.2
The Digital Aspects of the Project The THEOT project uses manuscript image sets from previous microfilming projects such as those of the Ethiopian Manuscript Microfilm Library,6 UNESCO,7 the Library of Congress,8 Hammerschmidt,9 Davies,10 and images of manuscripts microfilmed or digitized by libraries and museums through the first decade of this century. More recently, the digital collections include 6 EMML refers to the Ethiopian Manuscript Microfilm Library, the largest of all microfilm projects conducted in Ethiopia between 1973 and 1993. Some 9,600 manuscripts were microfilmed and a series of catalogues produced:  VOHD 20.1-3 (Wiesbaden: Steiner, 1973, 1977, 1999 We began to enter the domain of what we might call "higher order" digital humanities when we wrestled with questions of how to translate the methods of text-critical analysis into automated processes that the computer could perform.17 As we began to think through how to implement text-critical methodology, we crossed a threshold when we saw that the foundation of the entire enterprise needed to be built on the alignment into columns of every word from every manuscript. Therefore at the most basic level we identified text variation units not as passages, nor as phrases within sentences, but words. All words that share the same root were placed in the same column of the spreadsheet, and all 11 Besides being one of the three official partners of the EMML project, the Hill Museum and Manuscript Library has sponsored several other digitization projects of Geʿez materials. Among the most significant of these would be a joint project with Michael Gervers, University of Toronto, to digitize the collection at the monastery of Gunda words from all manuscripts receive a column. The most significant implication of this approach is that every detail of every manuscript is in the dataset and virtually all of it can be analyzed, aggregated, separated, or otherwise manipulated for analysis.18

Initial Findings
While there are a few disadvantages to the THEOT methods, the main advantage is that its statistical approach offers the ability to describe with precision both large patterns in the tradition as well as many minute phenomena.19 The initial results shared below will illustrate the benefits of this approach.

18
In dialogue with the rest of the team, Jost created programs in Google scripts that validate the formatting of the document holding the aligned texts, determine for each column the number of variants and assign a numerical value to each, profile the variants of each manuscript in each column against those assigned values, tally the percentage agreement of each manuscript with every other manuscript, create a statistical profile of every manuscript in terms of unique plusses and minuses, create "comma separated value" (csv) files for R statistical package from which dendrograms are generated, identify and list the distinctive readings of clusters of manuscripts (with the Geʿez text included for analysis), identify and profile the best representative manuscript of each cluster, profile the secondary and tertiary affiliations among the manuscripts (i.e., those readings that each manuscript shares with manuscripts outside its primary cluster), and analyze the diachronic developments of variants from cluster to cluster. Each of these scripts render data which, when interpreted, illuminate the various aspects of historical developments among the manuscripts. 19 Perhaps the greatest disadvantage involves the necessity of ignoring the nuancing of words (through prefixes, suffixes, and infixes) when aligning columns solely on the basis of shared roots. This decision obliterates all of the categories of evidence that Semitic languages encode in their morphological systems for nuancing a morpheme: singular and plural nouns, number in verbs, aspect of verbs, indicative and interrogative, etc. These categories are not insignificant, but the information gleaned from scripts that were programmed to distinguish among all of these options would deliver very little more about textual history than what is gleaned without these distinctions. The scripts could actually accommodate such information, but we judged that the programming challenges and additional time in formatting were prohibitive and, in the end, unnecessary for the project goals. Another significant disadvantage is the potential obfuscation of larger variation units and their impact on calculations of agreement with other witnesses. Even with the rich dataset provided by computer analysis, primary investigators must still examine all variations in context to provide a more accurate assessment.

2.1
Uniformity of Tradition Perhaps the single most important aspect about the Ethiopic Old Testament tradition is its remarkable uniformity.20 With a few exceptions at the periphery, the extant manuscripts of the last seven centuries reveal a tradition in which the vast majority of textual variation is played out in just 9 to 12 percent of the words in any given book. (Fig. 1 illustrates the high level of consanguinity found across fifteen books.) This makes it clear that the Ethiopic tradition descends from a single recension. Even if one can demonstrate that a cluster of manuscripts within the Ethiopic tradition shows evidence of influence from Arabic versions, for instance, it is questionable to refer to this cluster as a "recension" when the net impact of all variants is so minimal.
The level of textual variation among the extant Psalter manuscripts is very nearly a full order of magnitude smaller than with the rest of the books of the Ethiopic Old Testament that we have studied thus far. Most of that variation takes place within 1.4 percent of the words; all of it takes place within 2.5 percent. The reasons for this remarkable stability are undoubtedly found in the holding force of the chanting community. They did not just read the text; they memorized and recited the text on an almost daily basis, even when Geʿez ceased to be a commonly understood language in Ethiopia.

2.2
Transmission History Examining the small percentage of variation that does exist within the overall tradition, we have identified clusters of manuscripts that share similar readings. These clusters appear to delineate the main phases of Ethiopic textual history, because 1) the preponderance of manuscripts in a cluster comes from the same epoch,21 and 2) even when we use completely different manuscripts from book to book, the same clusters from the same timeframes appear.
The earliest attested manuscripts, from the fourteenth to the sixteenth centuries, contain a set of distinctive readings that set them off from other manuscripts as an identifiable "supercluster," i.e., a conglomeration of related witnesses made up of two or more clusters. (The boxes on the left side of Fig. 2   20 This is, perhaps, even more remarkable given the substantial number of variant recensions within the Arabic Old Testament tradition that permeated the ecclesial environment in Egypt, which was traditionally above Ethiopia in the church hierarchy. 21 Though it is always possible for later manuscripts to carry earlier texts, this is actually a rare phenomenon in the manuscripts studied thus far. It seems clear that it is not the antiquity of a text form that drives decisions about what specific text will be copied into a new manuscript but fidelity to the unfolding received text. assefa et al. Textus 29 (2020) 80-110 figure 1 Dendrograms of fifteen books of the Ethiopic Old Testament showing the general agreement in wording across the tradition encompass the early witnesses of Amos.22) Within that early supercluster, though, there are almost always at least two competing forms of the text. One of these clusters is often discernible as an attempt to improve the other. We refer to this particular form as a "Transitional Text," i.e., one that shares most of the readings of the earliest extant supercluster, while at the same time setting the stage for the next major cluster to emerge in the tradition.23 22 The dendrograms presented in Fig. 1 display agreement in wording. By focusing on agreement in variation (Fig. 2), the manuscript and cluster relationships become more sharply defined. 23 In other books, the Transitional Text would appear in a grouping similar to that which is found in the box farthest to the left in Fig. 2. Amos, however, lacks a Transitional Text.
Here the box delineates codices that descend from a poorly copied exemplar of the earliest attested text. Most of this group's distinctive readings are scribal blunders.
Textus 29 (2020) 80-110 figure 2 Dendrogram displaying agreement in variation in Amos Most of the newer manuscripts fall into a second supercluster. For Amos, the boxes on the right (Fig. 2) demarcate that group. Within this supercluster of younger witnesses, a major cluster, usually the largest, surfaces in the sixteenth and seventeenth centuries. We refer to this as the Standardized Text because the large number of manuscripts and their greater degree of uniformity appear to indicate a program of standardization sponsored by the combined authority and resources of church and state.24 It is often possible to show that this form of the text is based on a manuscript of the Transitional Text. The move from the earliest extant form of the text to the Standardized Text represents the major watershed of the extant Ethiopic Old Testament textual history.
Finally, a third major cluster emerges in the nineteenth and twentieth centuries as the form of the text we call the modern Textus Receptus. This form of the text is usually grounded on the Standardized Text. In fact, the modern Textus Receptus occasionally simply adopts the Standardized form of the text with only minimal changes. But usually this cluster is set off by a discernible assefa et al.
Textus 29 (2020) 80-110 set of distinctive readings.25 These three stages of the text, then, constitute the heart of the Ethiopic tradition and usually account for more than 85 percent of all the manuscripts.
Small clusters of manuscripts with a high degree of variation from the primary text forms described above are an indication of interesting and rare forms of the text. For instance, we identified two manuscripts that carried texts that were virtually identical to one another, but diverge significantly from all other clusters. 26 We were able to determine that the text had been translatedcomplete with versification and chapter summaries-from a Latin Vulgate edition.27 Martin Heide, who is completing a critical edition of the Ethiopic Jeremiah cycle, has isolated another small cluster of early manuscripts of Jeremiah that vary from the other earliest-attested manuscripts to a very high degree (around 45 percent).28 The cluster's differences reflect dependence on an Arabic Vorlage of Jeremiah that can be described as an amplified version of the Hebrew text. We have a similar case with a small cluster of later manuscripts of Ethiopic Daniel translated from an Arabic source containing both text and commentary. Cambridge Additional 1570 is a manuscript with many unique readings, produced in a unique historical setting, perhaps by a Jewish scribe. It shows some of the earliest evidence in the tradition of influence from the MT (more on this below). The point here is that the large data set clarifies not only the main trunk of the tradition, but also identifies movements at the margins of transmission history. We can identify them and analyze them, but we can also maintain perspective about their impact on the tradition.
At this point, we provide a few distinctive readings that demarcate the main stages of the textual history of the Ethiopic Old Testament. This is true of both 25 Note the box to the left of the Standardized Text witnesses in Fig. 2 the readings that separate the earlier texts from later ones and those changes distinctive of more narrowly defined clusters such as the Transitional Text, Standardized Text, modern Textus Receptus, and even smaller groupings. As a point of reference, Greek and Hebrew readings are provided since it is assumed that the earlier text forms preserve significant portions of the original translation from the Greek and the later text forms display changes toward the Hebrew, although almost always the latter are introduced from an Arabic source.29 Rarely do we note Greek and Hebrew variants, even when they might correspond better to the Ethiopic reading(s) being discussed.30 It is not our goal at the moment to determine origins of readings. Rather we seek instead to trace the history of transmission through those readings. This is not to dismiss questions of origin, for they are important for every new reading and new cluster, as a few of the observations below will substantiate.

2.2.1
Psalms31 We mentioned above that the manuscripts of Psalms have a degree of variation that is one-tenth of the usual variation in the other books of the Ethiopic Old Testament. Out of the sample passages, consisting of approximately fif-29 These historical developments have been observed in virtually every Old Testament book, but they have been most clearly documented in Löfgren's Daniel, i-liv, and in his textual commentary on pages 89-157. 30 For example, many of the newer Ethiopic readings align not just with Hebrew/Arabic but also with the Lucianic Greek text when it differs from LXXB text. Historically and linguistically, though, the evidence points to engagement with non-Greek sources making the more recent Ethiopic variants useless for textual criticism of the Greek Old Testament.  9, 12, 24, 30, 60, 99, 110, 161, 242, 243, 342, 343, 395, 2175;IES 74, 77, 837;EMML 2010EMML , 2064Leiden Or. 4734, 14617, 14664, 17086, 17087, 17088, 20531, 22026 indicate that Psalm 151 is "outside the number" (ወውፅእ፡ ውእቱ፡ እምኆልቍ; cf. καὶ ἔξωθεν τοῦ ἀριθμοῦ), which may have allowed for a different set of dynamics to be at play in its transmission. The psalm actually has three one-word "hotspots" in the text-the plus ወአድኀነኒ, "and he saved me," after "his angel" (cf. τὸν ἄγγελον αὐτοῦ) in v. 4, the plus ርኩሳን, "abominable," after "his idols" (cf. τοῖς εἰδώλοις αὐτοῦ) in v. 6, and the plus ለጎልያድ, "of Goliath," after "I cut off his head" (cf. ἀπεκεφάλισα αὐτὸν) at the end of v. 7-and one large hotspot of three strophes in the middle of v. 7, ወአንሰ፡ ነሣእኩ፡ ሠለሥተ፡ አዕባነ፡ እምእፍላግ። ወወጸፍክዎ፡ እንተ፡ ፍጽሙ። አሚሃ፡ ወድቀ፡ በኃይለ፡ እግዚአብሔር, "And I picked up three stones from the brook, and cast it to his head, when he fell down by the power of God." It is remarkable that these four hotspots each have their own textual history. By the end of the sixteenth century, the first variant had been fully adopted by the tradition. The second through the fourth variants appear in the seventeenth and eighteenth centuries, but their adoption each follows a different timetable with differing degrees of adoption by the manuscripts.

Song of Songs34
The Song of Songs is the most copied book of the Ethiopic Old Testament because it appears in virtually every Psalter, in biblical manuscripts containing the Solomon corpus, as one of the biblical texts in the Lectionary for Passion Week, and also in the Funeral Ritual. There are three distinct recensions of the Ethiopic Song of Songs.35 The common recension is the one that is most frequently copied and represents a fairly straightforward translation of the Greek text. This recension appears in the extant manuscripts of all periods, from the fourteenth century to the present. A second recension contains fully 128 additional strophes spread fairly evenly throughout the book. These strophes reflect extensive poetic, linguistic, aesthetic, and theological interests. The earliest manuscript of this recension is EMML 2064, from the fifteenth century. We know of only twenty codices that contain some or all of the strophes associated with this recension; a later hand often adds the strophes interlinearly. A 14, 22, 24, 30, 35, 44, 48, 60, 99, 110, 161, 242, 243, 342, 343, 395, 1055, 1060, 2175IES 74, 77, 837;EMML 2010EMML , 2064EMML , 2496EMML , 2497EMML , 2602Tanasee 103, 166;UNESCO 2.84;UNESCO 10.16;and Yale Ethiopic 18. 34 Steve Delamarter provides this survey of the evidence for Song of Songs based on his work as the Principal Investigator. About 850 Psalters have been probed to this point. The study is based on the full transcription of eighty manuscripts: thirty-one of the common edition, twenty-nine of the Scholars recension, and twenty of the Old Giyorgis recension. 35 Here it would be correct to employ the term "recension," indicating significant editorial differences.
-Deut 16:16 The older clusters read እግዚአብሔር፡ አምላክከ, "the Lord your God." In contrast, the newer ones have just እግዚአብሔር, "the Lord." There is an apparent tendency in older clusters to make the address "the Lord your God" consistent throughout the text. Cf. κυρίου τοῦ θεοῦ σου and ). The differentiation between the two main clusters of the older branch is complex, and it has not yet been possible to establish the precedence of one over the other. -Deut 15:15 Cluster 2 has ትግበር, "to do," which is in line with LXX: ἐγώ σοι ἐντέλλομαι ποιεῖν, but it is not found in Hebrew (  . Cluster 1 has ፍኖቶ፡ ለእግዚአብሔር, "way of the Lord," which preserves an early scribal error that has persisted through to the Standardized Text and the modern Textus Receptus. -Deut 32:2 Differences between the two clusters of earliest attested manuscripts often concern the choice of vocabulary. (See also the comments on 19:5 above.) Cluster 2 employs the specific, and perhaps more poetic ነጠብጣብ, "drops of moisture," whereas Cluster 1 uses the more generic ዝናም "rain." Cf. ὡς ὑετός and ).41 Cluster 1, on the other hand, has ወሠርዖሙ፡ ለአሕዛብ፡ በበደወሎሙ, "he established the borders of the nations." Differentiation between clusters of the newer manuscripts is straightforward. The Transitional Text has many differences from the Standardized Text and the modern Textus Receptus. Hence the Transitional Text of Deut 31:9 includes ለነገረ, "the word(s)" (= τὰ ῥήματα) found in the earliest attested clusters, but , which gives an imperative (a demand to Naomi), the Greek and the Ethiopic express the reaction or the feelings of Ruth (μὴ ἀπαντήσαι ἐμοὶ τοῦ καταλιπεῖν σε). The Hebrew could mean "Do not push me to leave you," but the Greek seems to say, "Let this not happen to me, to leave you …." The earliest Ethiopian manuscripts render the Greek ኢያብሐኒ, "may it not be permitted me [to leave you]." The Standardized Text and the modern Textus Receptus substitute ኢይትከሀለኒ, "it is not possible for me," which improves the reading. Such a change does not require influence from another version.
-Amos 9:7 The prophet lists foreign nations who, like Israel, have benefitted from God's salvific intervention. The earliest attested text refers to God's deliverance of "Israel from Egypt, the Ilofǝli from Cappadocia, and Syria from the pit (ግብ, changed to ግብጽ, "Egypt" in Aksum Zion, EMIP 654, GG 152c, and GG 203)." Cf. οὐ τὸν Ισραηλ ἀνήγαγον ἐκ γῆς Αἰγύπτου καὶ τοὺς ἀλλοφύλους ἐκ Καππαδοκίας καὶ τοὺς Σύρους ἐκ βόθρου. Here the Standardized Text is divided with one branch stating that God delivered the Israelites "from Egypt and from the Ilofǝli and Syria from Egypt"52 and the other declaring that God saved Israel "from Chaldea and from Mǝkaran and from Egypt," with "Chaldea" and "Mǝkaran" resulting from an incorrectly read Arabic text.53 The modern Textus Receptus also preserves two different readings: with Israel rescued "from Chaldea and from Mǝkaran and from Egypt and from Ilofǝli" in one cluster54 and "from Chaldea and from Mǝkaran and from Syria and from Egypt" in another.55 EMML 6930, another member of this group, conflates both streams. The cluster more directly influenced by Hebrew retains the first two pairings found in the earliest attested text but alters the last to "Aram from Mǝkaran"; cf. .
-Obad 9 The older manuscripts unanimously read እምደወለ, "from the territory of," while six out of eight (75 percent) of the younger manuscripts substitute እምብሔረ, "from the land of." The Greek has ὄρους, "mountains" but translators often mistook this word as a form of ὅριον. (See Deut 32:8 above for a similar error.) The shift from ደወል to ብሔር is probably related to diachronic changes in the native language. According to Leslau, there remains no Amharic cognate of ደወል meaning "territory" in common use. the Hebrew. Perhaps the scribe was influenced by the very final words of Obadiah, ወይከውን፡ መንግሥተ፡ ለእግዚአብሔር, "and there will be the kingdom of God," or by an external source. The following two variants show a mistake in the earliest manuscripts followed by an attempt to correct it, which eventually did not win the day. Instead, the mistake persists through the Standardized Text form and into the modern Textus Receptus: -Obad 1 The earliest attested manuscripts have ወፈነወ፡ ማዕገተ, "and he sent an ambush." This curious rendering follows the Greek περιοχὴν … ἐξαπέστειλεν rather than the Hebrew , for "ambush" is simply a translation error, the translator having misread περιοχήν as derived from περιέχω. Beginning in the seventeenth century, several manuscripts provide a correction, reading ወፈነወ፡ ሐዋርያ, "and he sent a messenger,"62 and several others conflate the readings ወፈነወ፡ ሐዋርያ፡ ማዕገተ፡ ላዕለ፡ አሕዛብ, "and he sent a messenger, an ambush upon the nations."63 Since the corrected reading "and he sent a messenger" is consistent with Greek and Hebrew, one would expect it to carry the day. However, the majority of manuscripts preserve the earlier reading. The two manuscripts that have a Latin Vulgate Vorlage correct the reading to "and he sent a messenger to the nations."64 -Obad 20 The oldest manuscripts unanimously read በሐውርተ፡ እስራኤል, "in the lands/regions of Israel." Since both Greek and Hebrew have "Jerusalem" (ἡ μετοικεσία Ιερουσαλημ and ‫ְי‬ ‫ר‬ ‫וּ‬ ‫שׁ‬ ָ ‫ַ֖ל‬ ‫םִ‬ ‫ְו‬ ‫ָג‬ ‫ֻ֥ל‬ ‫ת‬ ), it is not clear why the oldest manuscripts have "Israel." It is worth noting that all manuscripts refer to Israel several words earlier in the verse, which may explain the corresponding "Israel" (instead of "Jerusalem") here. Several manuscripts provide a correction by replacing እስራኤል, "Israel" with ኢየሩሳሌም, "Jerusalem."65 However, as was the case in the previous variant, these corrected texts did not attract a sufficient following, and so it was the earlier reading "Israel" that persisted through the Standardized Text to the modern Textus Receptus. The following variants help characterize the modern Textus Receptus:66 -Obad 4 All manuscripts attest the reading ማእከለ፡ ከዋክብት, "among the stars." Among the collated witnesses, only one seventeenth century manuscript,67 . "Jerusalem" avoids redundancy while providing a satisfactory parallel to "Zion," which appears in the newly added phrase noted above.
-Obad 17b All but three of the manuscripts examined read ወይከውን፡ ቅዱሰ, "and it will be holy." Cf. καὶ ἔσται ἅγιον and . Three of the four modern Textus Receptus manuscripts70 add ለእግዚአብሔር, "to God," creating an additional parallel to the recently added phrase. (See notes on Obad 17 above.)

2.2.7
Jonah71 Ethiopic Jonah is transmitted in the typical patterns described above as well as in a text type that is a translation of the Latin Vulgate.72 The discussion below focuses on the typical pattern as the translation of the Latin Vulgate clearly was produced by a different mechanism than was the rest of the extant manuscript tradition.
The earliest attested text73 and the earliest form of the "Transitional Text"74 differ in a number of places that can be divided into three categories: 1) readings found in the earliest attested text are closer to the Greek than those of the Transitional Text; 2) readings preserved in the Transitional Text are closer to the Greek than the earliest attested readings; and 3) readings not so easily placed in either category. Examples from each category follow: -Jonah 2:4 The second-person singular verb form in the Transitional Text, ወወረውከኒ, "and you cast me," is closer to the Greek ἀπέρριψάς με than is the ) as እምፍጥረተ፡ ዓለም, "from the creation of the world." The Transitional Text lacks an equivalent.
-Jonah 2:8 The Transitional Text orders the two clauses of the strophe in the same order as the Greek, but the earliest attested text has the two clauses in the opposite order. However, the earliest attested text translates ἀπ᾽ ἐμοῦ with በላዕሌየ, "upon me," whereas the Transitional Text does not offer a translation. Cf. ἐν τῷ ἐκλείπειν ἀπ᾽ ἐμοῦ τὴν ψυχήν μου τοῦ κυρίου ἐμνήσθην, καὶ ἔλθοι πρὸς σὲ ἡ προσευχή μου εἰς ναὸν ἅγιόν σου and There is a strong divide between these two earlier text types and the two text types that come after them. The Standardized Text type is a revision of one branch of the earliest attested text (Earliest Attested 2 rather than Earliest Attested 1). The differences between the Standardized Text and its antecedents are often due to cross-pollination from other sources such as liturgical or biblical texts, other text types of Ethiopic Jonah, or with an external source such as an Arabic translation of the Syriac Peshitta.
-An example of cross-pollination with an external source is found in Jonah 1:9. The earliest attested text has ገብረ፡ እግዚአብሔር, "servant of God," which is a translation of LXX δοῦλος κυρίου. The Standardized Text has ዕብራዊ, "Hebrew," which is a translation of an Arabic source translated from a Syriac text based on Hebrew ( ‫ִﬠ‬
-Jonah 2:4-10 The reading of the earliest attested text is largely replaced in the Standardized Text with the version of these verses found in the Seventh Biblical Canticle as transmitted in the Ethiopic Psalter. -Jonah 4:2 The Standardized Text has a plus of ወጻድቅ, "and righteous," after "much mercy" that is absent in the earlier text types. (Cf. πολυέλεος and .) This plus is likely due to liturgical influence from other biblical texts, such as Ps 85:15. The modern Textus Receptus revises the Standardized Text. Its revisions often blend or conflate elements of one strand of the earliest attested tradition and the Standardized Text. For example, the Textus Receptus conflates the two readings of Jonah 1:9 described above to read ገብረ፡ እግዚአብሔር፡ አነ፡ ዕብራዊ, "I am a servant of God, a Hebrew." In addition, the modern Textus Receptus is characterized by longer readings than those found in the earlier text types. The twentieth-century witnesses attest expanded readings in sixteen of the verses of Jonah.
Textus 29 (2020) 80-110 2.2.8 Haggai75 In Ethiopic Haggai, the distinctions between the two earliest attested forms of the text cover fifteen points of variation. Several of these are easily identified as scribal errors (such as the confusion between አመ, "when," and ከመ, "that," in 2:16 or the repetition of ነፍስ, "soul," in 2:13), but there are a few more telling variants worth noting here: -Hag 1:2 The earliest attested text reads ኢኮነ, "was not," whereas the Transitional Text has ኢበጽሐ, "had not arrived." Cf. οὐχ ἥκει and -Hag 2:10 ኮነ፡ ቃለ፡ እግዚአብሔር፡ ኀበ፡ ኃጌ፡ ነቢይ፡ ወይቤሎ, "The word of the Lord came to Haggai the prophet, and he said to him," found in the earlier witnesses differs from the Transitional Text's ነበበ፡ እግዚአብሔር፡ በእደ፡ ሐጌ፡ ነቢይ፡ ወይቤሎ, "God spoke through Haggai the prophet, and he said to him." Cf. ἐγένετο λόγος κυρίου πρὸς Αγγαιον τὸν προφήτην λέγων and Perhaps the most distinguishing feature in the earliest attested text, however, is one with great stemmatic importance for the later transmission of Haggai: the appearance of an angel (መልአክ) in the middle of one of Haggai's oracles. As the angel only appears in locations where Haggai's name is mentioned in LXX (2:13a and 2:20b), its presence in the text is most easily explained as a result of misreading Αγγαιος, "Haggai" for ἄγγελος, "angel." Regardless, its awkward narrative locale has resulted in two different paths of transmission beyond the earliest attested texts. First, the Transitional Text refers neither to the angel nor to Haggai. This minus remains in later texts that build on the Transitional Text. Second, the Standardized Text and modern Textus Receptus embrace the presence of the angel, and indeed further explain in v. 20 that it was the Lord (እግዚአብሔር) who spoke to the angel. Interestingly, apart from the Latinized Version and a single interlinear correction in BN Abbadie 35, Haggai's name is not found among the manuscripts studied, even those manuscripts exhibiting influence from the MT.
The textual history of Ethiopic Haggai, from its earliest attested form through the Standardized Text and into the modern Textus Receptus, is succinctly displayed by the dating of Haggai's four oracles account for four of the sixteen points of variance that separate these three forms from one another. Interestingly the four oracles are all dated the same in the Ethiopic, and they read thus: -The earliest attested text reads ተሱዑ፡ ለጽልመት፡ ታስዕ, "the ninth of the ninth darkness (i.e., month)." -The Standardized Text reads ፳ወረቡዑ፡ ለታስዕ፡, "the twenty-fourth of the ninth." -The modern Textus Receptus conflates these two readings to read ተሱዑ፡ ለጽልመት፡ አመ፡ ፳ወ፬ለታስዕ, "the ninth darkness (i.e., month) on the twentyfourth of the ninth." Each of these clusters stands at odds with LXX (τῷ ἑβδόμῳ μηνὶ μιᾷ καὶ εἰκάδι τοῦ μηνός) and MT ( ) at 2:1, although the latter two more closely fit the Greek and Hebrew at 2:10, 18, and 20.

2.3
Scribal Idiosyncrasy Despite the remarkable uniformity of the general tradition, we can identify and describe the degree to which individual scribes introduced idiosyncratic readings into their copies. By organizing the spreadsheet of the transcription data so that words of the same root appear in a single column, each column bears witness to one of three phenomena: 1) complete uniformity, where every manuscript has the same root word; 2) cluster variation, where one group of manuscripts has a reading that others do not; or 3) unique readings, where one manuscript varies from all the rest. For example, an excerpt of the transcriptions of Amos 9:6a below shows that there is: 1) complete uniformity with ዘየሐንጽ, "the one building," which appears in one form or another in all collated witnesses; 2) cluster variation where ዕርገቶ, "his ascension" appears only in the Latinized Text (W17 and W18) against ጽርሖ, "his upper chamber," which is found in all the other clusters; and 3) a unique reading in EMML 1768 (W11) with እስከ, "as far as" replacing the otherwise ubiquitous ውስተ, "in." Assuming that unique readings such as the one found in Amos 9:6 by EMML 1768 attest scribal idiosyncrasy rather than readings of other as yet unidentified manuscript clusters, a typical scribe creates fewer than ten idiosyncratic readings within the sampled text, but a small number have more than twenty or thirty, up to two and three standard deviations off the norm. By aggregat-Textus 29 (2020) 80-110 ing these statistics, we achieve a profile of the entire cohort of scribes whose manuscripts have been collated. This body of data provides an interesting insight into patterns of scribal behavior. With it, one can detect, for instance, if there may be epochs or locations in which scribal idiosyncrasy may have been allowed or prohibited. Judging by the mean, manuscripts produced at the monastery of Gunda Gunde, for instance, show a remarkable lack of idiosyncrasy. This sort of profile of the behavior of the cohort of scribes is possible only because of the statistical approach to the work and the fact that no piece of data is simply excluded from consideration because it is unique.

2.4
Contact with the MT76 Opportunities for Ethiopian contact with the Hebrew Bible existed before the country's conversion to Christianity in the fourth century CE. As a major conduit for international trade, Ethiopians regularly encountered a wide variety of people until the Aksumites retreated away from the coast into the highlands. Even as the nation became more isolated geographically and economically, religiously it remained tethered to other locales, including a notable, although perhaps not continual, presence in Jerusalem. And occasionally Jews traveled to Ethiopia, sometimes having an impact on the royal family and therefore their religion.
Nevertheless, it is only late in Ethiopia's literary history when influence from the Hebrew Bible appears. Its greatest impact is channeled through the "daughter versions." For many biblical books, an Arabic translation of the Syriac, which itself descends from a text close to the MT, influenced the creation of the Standardized Text. Yet, there are a few places where translations (e.g., Prov 10:4) and transliterations (e.g., Neh 11:17 and Dan 11:30) of Hebrew rather than Arabic betray direct Hebrew influence. This evidence fits what we find stated in colophons associated with the Standardized Text, namely that the reviser(s) employed Hebrew along with Arabic and Coptic sources.77 The later, nineteenth century revision against the Latin Vulgate also produced a text much closer to the Masoretic Text than its predecessors.
More concrete contact with the Masoretic Text does appear, although it is sporadic up until modern times. One of the earliest examples comes from BN éth. 7, a fourteenth century manuscript of Job and Daniel. In the margins of that manuscript, a later reader (fifteenth century?) added notes where a certain Marqoryos observed that the codex's text deviated from the Hebrew. 76 See, further, E. Ullendorff, "Hebrew Elements in the Ethiopic Old Testament," JSAI 9 (1987) The enduring value of the project, though, will be in the digital archive of data we generate and the tools developed for analysis and visualization. The data produced so far have advanced our understanding of the transmission history of the Bible in Ethiopia. First, the more precise picture allows us to map the historical developments of the text with a level of granularity virtually unexplored in earlier studies. In addition to the general distinction between older and younger texts made in those studies, smaller clusters have also been identified, including a fascinating chapter in Ethiopia's transmission history: a thorough revision based on the Latin Sixtine Vulgate. The greater precision further highlights common catalysts for textual change in Ethiopia such as intertextuality, influence from external textual traditions (primarily Arabic and Hebrew), and scribal error. It also brings to the fore the limited number of stylistic improvements in a tradition originating with a generally word-for-word rendering of the Greek Vorlage with syntactical features unnatural in Geʿez.80 Second, the more complete picture the project provides is just as important. For various reasons, it has been standard practice in modern, Western scholarship to produce editions of Ethiopic books of the Bible in which the evidence of the manuscripts after the seventeenth century has been systematically excluded. But the goal of textual criticism is understanding the history of textual transmission, not just reconstructing an early exemplar. Further, Ethiopian Orthodox Christianity-like other ancient and non-Western expressions of Christianity-finds the truest and best form of its authoritative texts not in the ipsissima verba of a theoretical text form that lies at the origins of the tradition, but in the unfolding text continuously being refined by the community generation after generation. Thus, arbitrarily omitting centuries of data falls far short of the text-critical goal.
It was typical of colonialism that it exploited the cultural and physical resources of dominated cultures for its own purposes and ignored or-worsesuppressed the use of those resources by the cultures for their own purposes.81