jarodjacobs/ethiopic_psalters_ml

Computational Codicology: An Exploration of Ethiopian Psalters In his project “The Social Lives of Ethiopian Psalters,” Steve Delamarter has identified over 110 features of codicology and scribal practice and book culture which are executed in differing and various ways by scribes, craftsmen, and book users. He selected 29 of these for further study and then created typologies of those variation, assigned numbers to each type of the variation and built a database (containing more than 100 fields) to receive data from Psalters. He has analyzed over 700 Psalters—ranging from the earliest extant ones of the fourteenth century up through the 20th century—and collected data regarding the variations in each. Each of the Psalters is dated; over 200 of the Psalters can be located in terms of the area of Ethiopia in which they were produced. The data produced from the project promises to open many windows on aspects of Ethiopian scribal and book culture. But this will only happen if the data is analyzed well. This presentation will focus on the ways in which we have applied specialized statistical analyses, such as cluster and random forest analysis, to the data. The raw data already suggest some fascinating stories; but a thorough statistical analysis is need to draw firm conclusions. Codicologists have already noted a few features that have changed over time; this dataset promises to greatly increase those findings. But perhaps even more interesting is the potential of identifying features that differ depending on where the manuscript was produced. Of all the manuscripts the place of origin is recorded and this information is also taken into account in the analysis. It is to be expected that a comprehensive analysis of both chronological and geographical data will lead to a much better understanding of Ethiopic scribal culture. # ethiopic_psalters_ml Computational Codicology: An Exploration of Ethiopian Psalters In his project “The Social Lives of Ethiopian Psalters,” Steve Delamarter has identified over 110 features of codicology and scribal practice and book culture which are executed in differing and various ways by scribes, craftsmen, and book users. He selected 29 of these for further study and then created typologies of those variation, assigned numbers to each type of the variation and built a database (containing more than 100 fields) to receive data from Psalters. He has analyzed over 700 Psalters—ranging from the earliest extant ones of the fourteenth century up through the 20th century—and collected data regarding the variations in each. Each of the Psalters is dated; over 200 of the Psalters can be located in terms of the area of Ethiopia in which they were produced. The data produced from the project promises to open many windows on aspects of Ethiopian scribal and book culture. But this will only happen if the data is analyzed well. This presentation will focus on the ways in which we have applied specialized statistical analyses, such as cluster and random forest analysis, to the data. The raw data already suggest some fascinating stories; but a thorough statistical analysis is need to draw firm conclusions. Codicologists have already noted a few features that have changed over time; this dataset promises to greatly increase those findings. But perhaps even more interesting is the potential of identifying features that differ depending on where the manuscript was produced. Of all the manuscripts the place of origin is recorded and this information is also taken into account in the analysis. It is to be expected that a comprehensive analysis of both chronological and geographical data will lead to a much better understanding of Ethiopic scribal culture.