Training documents and materials for the DAA 2022, held at Uppsala University in Sweden 1-5 August
View this repository on GitHubglow-gh/daa
This course block introduces the idea of using networks to explore relationships between words rather than people, in an approach called Lexical Network Analysis (LNA). This module will introduce you to the concept of word co-occurrence, the probabilistic measurement called Pointwise Mutual Information (PMI), and the analysis of networks built from co-occurrence data in Gephi.
The course will begin with a lecture to explain the key elements of LNA, and how it differs from SNA. During the module, we will:
Suggested reading: Svärd, S., Alstola, T., Jauhiainen, H., Sahala, A. and Lindén, K. (2020) ‘Fear in Akkadian Texts: New Digital Perspectives on Lexical Semantics’, in S.-W. Hsu and J. Llop-Raduà (eds) The Expression of Emotions in Ancient Egypt and Mesopotamia. Leiden: Brill (Culture and History of the Ancient Near East, 116), pp. 470–502. Available at: https://doi.org/10.1163/9789004430761, and as a pdf in Github.
As with the SNA module, you can refer to this list of technical terms for Network Analysis for jargon and complex concepts.
See also further readings, if you want to learn more.
Ensure the following links work on your computer:
If you have any issue with the tasks prior to the session, please send an email to Ellie (eleanor.bennett@helsinki.fi).
There will be three parts to this module, each about an hour long. After each module we will have a short 5 minute break so you can stretch your legs.
We will begin with a short lecture (also available as a pdf). This will introduce the concept of building networks based on relationships between words. This will include a brief look at word co-occurrence methods, and some examples of how LNA has been used in Assyriological and other ancient texts.
We will then look at the worfklow of LNA to understand the steps needed in building a lexical network. This will include:
Finally, we will talk about what we need to do when we want to find the PMI co-occurences for more than three Akkadian words.
In this section, you will begin exploring a pre-prepared lexical network. In the first half hour, you will split into groups and play with the layout of the network. You will be exploring which layout algorithm seems to display the network best. You will also be changing the size and colour of the nodes based on network measurements. This will help you grapple with how lexical networks based on PMI are different from social networks.
In the second half hour, your group will be assigned a word to look at in your network. You will use the network to find patterns surrounding the assigned word, and then learn how Korp can help explain the patterns. An example complex search based on the PMI parameters can be found here.
In the final section, we will compare your network with two others:
We will end with a plenary discussion about the pros, cons, and potential for this method in Ancient Near Eastern studies.
Co-occurrence networks in historical documents is a relatively new development of Network Analysis as a field. However, if you are interested in using this method for future research, this is a selection of articles and books we have found useful. They are included in the zip file.