Data Science and the Bible: Rediscovering the History of the Christian Bible in Ethiopia
October 2021 | Vol. 9.10
By Steve Delamarter and Daniel Assefa
For those in the Mediterranean basin, Ethiopia represented a distant and exotic land. Ethiopia was many miles from Rome, Jerusalem, and Alexandria, but much of the barrier between Ethiopia and the rest of the world was the elevation. Rising abruptly from the Red Sea, the highlands of Ethiopia tower to a height of two miles and more. The region is marked by rough, often impenetrable, terrain but also boasted fertile areas that supported a significant state centered in the city of Axum.
From Rome to Ethiopia. The figure combines an image from Google Earth and a picture of the Seimen mountains in the Ethiopian Highlands, the birthplace of Christianity in Ethiopia, by Hulivili, CC By 2.0.
Decades before the Roman Empire adopted Christianity in the early third century, Syria, Armenia, and Ethiopia had all made Christianity the religion of the state. As a first priority, each began the arduous task of translating the Christian scriptures into their languages.
We know that the work was begun in Ethiopia because of stone inscriptions in and around Axum, the original capital of Christian Ethiopia. Several were erected from the fourth to the sixth century and some contain small excerpts from biblical texts, translated into Ge’ez, the language of Ethiopia.
The Ezana Inscription. Photo by Sailko / Wikimedia, CC BY 3.0
In spite of this early evidence we know almost nothing about the first millenium of Bible translation and copying in Ethiopia. Very few manuscripts are clearly dated to this early period; the vast majority are from the fourteenth century onwards, with few from the fourteenth to the sixteenth centuries, then in ever-increasing numbers.
Notable exceptions are the so-called Abba Garima Gospels. These remarkable manuscripts reside in the Endā Abbā Garimā Monastery in Northern Ethiopia, not far from Axum, and may date to the sixth century CE or even earlier. They are remarkable not only for their texts and ancient script but also for the richness of their illuminations.
Abba Garima Gospel 1, images 9 and 10, combined. This manuscript contains the four gospels, showing here two of the Eusebian canon tables. Image by Ethiopian Heritage Fund via Wikimedia.
The Ethiopic Bible is translated from Greek, both the New Testament—originally written in Greek, but also in the Old Testament—originally written in Hebrew. The Bible in Ethiopia and many other places is thus among the early Christian translations known as the “daughter versions of the Septuagint.”
Time and changing language invariably builds a wall between believing communities and their sacred texts. The vast majority of Christians in Ethiopia do not read or speak the old language of Ge’ez. Most use Amharic, the common language of modern Ethiopia, or other local languages. Despite this, manuscripts of the Ge’ez Bible are treated with profound respect in churches and monasteries. They are sacred objects in themselves; during part of the mass, the priests work their way through the congregation holding a copy of the Four Gospels and the members of the church touch it to their foreheads and kiss it.
The sociology of the Christian tradition has generally been conservative, insisting that the language of sacred text and liturgy remain in the old sacred languages into which they were first translated—whether Ge’ez, Armenian, Syriac, or Slavonic. The price was that only the deep insiders of the community—priests and clerics—continued to master the language of sacred text and liturgy. Laypeople thus encounter liturgy and sacred text in a remarkable, but odd, way. They are able to recite huge texts, but they experience them as collections of syllables. This was a fundamental characteristic throughout the medieval world, and not just the Latin-speaking parts.
Given this concession from the laypeople in the tradition, one might imagine that the original translation would never need revision. This is not the case. The Ge’ez Bible has been continuously copied and recopied across the better part of two millennia. And even though we have no copies of most of the book beyond the last seven hundred years, the manuscripts that do survive reveal an interesting story: The Bible in the old language of Ge’ez has continued to develop and change—only very slowly.
The history of the Ge’ez bible in Ethiopia in the last seven centuries is no small matter. To understand how the text of the Ethiopic book of Ruth changed over time requires data science. The two authors of this article work together with colleagues (Garry Jost, Ralph Lee, Curt Niccum, James Prather, and Brent Reeves) in a project called THEOT (the Textual History of the Ethiopic Old Testament) to reconstruct the textual history of the Bible in Ethiopia.
We first identified a minimum of thirty copies of the Ethiopic Book of Ruth evenly distributed across time; 4 or 5 manuscripts for each century from the fourteenth to the twentieth. Then we created very accurate transcriptions of each to make comparisons between and among them. We can then line them up for comparison.
Thirty-seven copies of Ruth 1:2a, transcribed and aligned into columns (a dot means no word is present).
In English, the first four words say, “And the name of the man was Abimelech.” The last three words say, “And the name of his wife was Naomi.”
In this sample of 37 manuscripts, we can glimpse of the typical changes that go on in this book culture, in seven hundred years of transmitting the book of Ruth. The big patterns of shared variation center on the presence or absence of the last three words, “And the name of his wife was Naomi.” Twenty of the thirty-seven manuscripts do not have these words. It is not possible to attribute this lacuna to individual idiosyncrasy.
Are there other places in the book of Ruth where these twenty manuscripts have unique readings? The only way to answer is for the computer to count up all the shared variations within all the manuscripts. Now we are able to compare the patterns of words present or absent from manuscript to manuscript.
Data Sheet showing the percentage agreement between every manuscript and every other manuscript.
Each cell is the intersection of two manuscripts; the percentage is the agreement between those two manuscripts. The intersection between W15 and W15 is 100% because the manuscript always agrees with itself. But W15 and W10 share only 79.6% of the words. W15 and W14 share 94.5% of the words. These are clearly much more closely connected.
It is not easy for our brains to make sense of dense tables filled with numbers. Fortunately, we can render all of those numbers into a visualization that is much easier to see the patterns.
A dendrogram of the relationships of the manuscripts in our study of the Book of Ruth.
Now we can start to see more patterns based on the “Disagreement Scale,” the degree to which the manuscripts are different from one another. The most important conclusion is at the top of the chart. This figure actually shows only the bottom 20% of the chart. If we could see the entire chart, it would be clearer to what degree the manuscripts agree: the top 80% of the chart shows complete uniformity in all the manuscripts in the tradition. At the bottom 20% of the dendrogram we see all the differences in the manuscripts. The copies of the book of Ruth are remarkably uniform in the Ethiopic tradition: most share a uniformity of about 92% and more. Across seven centuries, this is remarkable.
But within this tradition is everything the same? With only four exceptions, the manuscripts of the Book of Ruth fall into one or the other of two large groups. All of the oldest manuscripts—from the fourteenth to the sixteenth—fall in the supercluster to the right, and the ones in the supercluster in the center left, fall in the seventeenth to the twentieth centuries. We have just identified the major moment of change in the extant Ethiopic manuscript tradition. Something happened in Ethiopia in the sixteenth century that brought about the biggest changes in the history of the extant Ethiopic Biblical text.
With a bit more study, we see sub-clusters on the right. Additionally, we see that the four manuscripts on the very left stand apart. When we bring all of these insights together we can assign cluster names and basic chronology.
Dendrogram of the Ethiopic Book of Ruth, with cluster identifications.
If we provide the computer with this breakdown of manuscript clusters, we can begin to search for the specific places in the book of Ruth where the readings of the clusters diverge from one another. Where the clusters share distinctive readings that proves our emerging theory of the history of the text.
The earliest manuscripts omit the phrase “and the name of his wife was Naomi.” The later texts have supplied the missing phrase, in conformity with the Masoretic Hebrew text (MT) and the Greek text (LXX). This seems to be a case where the phrase dropped out in the very earliest stages of the text in Ethiopia and which was restored over a millennium later in the seventeenth century.
The earliest manuscripts omit “she heard that.” According to these readings, the reader would not know how she learns about the end of famine in Israel. The later version renders the correct translation of the Hebrew.
The later version has added “and he drank”, a reading present in the MT, but absent from the earliest attested manuscripts, as well as from the LXX.
In the fourteenth through the sixteenth century we have the Earliest Attested Text. The variations among these early manuscripts is not insubstantial, ranging from two to six percent between any two manuscripts. But in the seventeenth century, the most substantial changes in text take place. We call it the rise of the Standardized textform.
There are several dozen single words and small phrases where this textform differs from the Earliest Attested textform, averaging about four percent. However, among the Standardized manuscripts copied from the seventeenth to the twentieth century the variations are three percent and less. In most other books of the Ethiopic Old Testament, there is one more textform that arises in the nineteenth and twentieth centuries, the Modern Textus Receptus. In the book of Ruth, these two are identical, which is to say that the Modern Textus Receptus simply passes on the form of the book as developed in the era of the Standardized text.
Through data science we are able to see the various texts produced in Ethiopia for the Book of Ruth. In the next figure, which our colleague Garry Jost created, displays the clusters and where they differ. We can also display the full history of the readings of the manuscripts, brought together in their clusters, and showing all their readings.
Full transcriptions of the manuscripts, arranged by cluster.
In the process of this work, we glimpse the Ethiopian community working diligently across time to pass on the sacred text delivered to them. Indeed, Ethiopia has had not one, but two remarkable book cultures, one Christian and one Islamic. Both communities invested huge resources and human capital into the production of their sacred texts. Every scribe had to master a substantial body of specialized knowledge. The finished products are a wonder to behold, even as the ravages of time threaten to reduce them to fragments.
One of the most amazing things about Ethiopia and the Bible is that the ancient skills of parchment bookmaking are still alive today. But it is under stress. The arrival of the printing press in the early twentieth century gutted demand for handmade books. And the modernization of Ethiopian culture pressures the ancient book culture from all sides. Time is running out. The more esoteric skills and knowledge of scribes is no longer known or practiced. We may have only another generation or two, before the book culture is so compromised that it no longer provides a living connection to the past.
But the manuscripts themselves remain. They contain all of the marks left on them by craftsmen, scribes, artists, and readers. Thousands of moments from Ethiopia’s history are trapped inside. To experience these, all one needs to do is lean forward into the book and smell the incense that soaked into the parchment, or count the wax drops on the folios where the candles were held over the books. In so doing, you are transported into the living past of Ethiopia’s Christian community. And alongside our own senses, today we have the tools of the digital humanities and data science to help us reconstruct those stories.
How to cite this article:
Delamarter, S. and D. Assefa. 2021. “Data Science and the Bible: Rediscovering the History of the Christian Bible in Ethiopia.” The Ancient Near East Today 9.10. Accessed at: https://anetoday.org/christian-bible-in-ethiopia/.
Want to learn more?
Post a comment