Schedule | Assignments | Fine Print | References

Independent study, fall 2016. Department of History and Art History, George Mason University. Meets every other Wednesday at 1 p.m. for discussion; work sessions on alternate weeks as necessary. Instructor: Lincoln Mullen <lmullen@gmu.edu>. Office: Research Hall 457.

This independent study is an advanced course in the theory and practice of text analysis for historians. You will read current research in digital history and cognate fields. The aim is to learn the methods of text analysis which are most likely to produce insights useful for historical interpretation. You will work primarily in the R programming language and with common Unix-style utilities, performing an analysis using each method we discuss week by week. By the end of the semester, you will write a paper which uses text analysis of a corpus to make a historical argument or interpretation.

Schedule

Week 1 (August 31): Introduction

• Matthew K. Gold et al., “Forum: Text Analysis at Scale,” in Debates in the Digital Humanities 2016 (University of Minnesota Press, 2016), 525–568, http://dhdebates.gc.cuny.edu/debates/text/93.
• Stephen Robertson, “The Differences Between Digital Humanities and Digital History,” in Debates in the Digital Humanities 2016, ed. Matthew K. Gold and Lauren F. Klein (University of Minnesota Press, 2016), 289–307, http://dhdebates.gc.cuny.edu/debates/text/76.
• Dan Cohen, “Searching for the Victorians,” October 4, 2010, http://www.dancohen.org/2010/10/04/searching-for-the-victorians/.
• Ted Underwood, “Seven Ways Humanists Are Using Computers to Understand Text. The Stone and the Shell,” June 4, 2015, https://tedunderwood.com/2015/06/04/seven-ways-humanists-are-using-computers-to-understand-text/.
• Matthew L. Jockers, Text Analysis with R for Students of Literature (Springer, 2014), http://link.springer.com/10.1007/978-3-319-03164-4, chs. 1–2.
• Taylor Arnold and Lauren Tilton, Humanities Data in R (Springer, 2015), http://link.springer.com/10.1007/978-3-319-20702-5, chs. 1–2.

Practicum: Before our first meeting, install R and the RStudio Desktop IDE (you may wish to install the preview version to get notebook support). Start to become familiar with the basics of R as described in the introductory chapters of either Jockers or Arnold and Tilton. You should also become start to become familiar with the basics of the Unix-style command line (see Shotts, Linux Command Line, as a reference).

Week 2 (September 7): Working with corpus metadata

• Stephen Robertson, “Searching for Anglo-American Digital Legal History,” Law and History Review 34, no. 4 (August 2016), doi:10.1017/S0738248016000389.
• Tim Hitchcock and William J. Turkel, “The Old Bailey Proceedings, 1674–1913: Text Mining for Evidence of Court Behavior,” Law and History Review 34, no. 4 (August 2016): 1–27, doi:10.1017/S0738248016000304.
• Benjamin Schmidt, “Women in the Libraries. Sapping Attention,” May 8, 2012, http://sappingattention.blogspot.com/2012/05/women-in-libraries.html.
• Julia Flanders and Fotis Jannidis, “Data Modeling,” in A New Companion to the Digital Humanities, ed. Susan Schreibman, Ray Siemens, and John Unsworth (Wiley Blackwell, 2016), 229–37.
• Stéfan Sinclair and Geoffrey Rockwell, “Text Analysis and Visualization: Making Meaning Count,” in A New Companion to the Digital Humanities, ed. Susan Schreibman, Ray Siemens, and John Unsworth (Wiley Blackwell, 2016), 274–90.

Practicum: Download at least one of the provided corpora. Extract the metadata into tables. Do an exploratory data analysis of the corpus metadata, paying special attention to the question of which texts actually belong in the corpus. See the R Markdown documentation for help getting started with your first notebook. See Jenny Bryan, Happy Git and GitHub for the useR, for guidance using Git/GitHub.

Week 3 (September 21): Vector space models

• A. W. B. Simpson, “The Rise and Fall of the Legal Treatise: Legal Principles and the Forms of Legal Literature,” The University of Chicago Law Review 48, no. 3 (1981): 632–679, doi:10.2307/1599330.
• Shawn Graham, Ian Milligan, and Scott Weingart, Exploring Big Historical Data: The Historian’s Macroscope (Imperial College Press, 2015), ch. 3.
• Arnold and Tilton, Humanities Data in R, chs. 9-10.
• Michael Witmore, “Text: A Massively Addressable Object,” in Debates in the Digital Humanities 2012 (University of Minnesota Press, 2012), http://dhdebates.gc.cuny.edu/debates/text/28.
• Benjamin Schmidt, “Age Cohort and Vocabulary Use. Sapping Attention,” April 11, 2011, http://sappingattention.blogspot.com/2011/04/age-cohort-and-vocabulary-use.html.
• Jure Leskovec, Anand Rajaraman, and Jeff Ullman, Mining of Massive Datasets, 2nd ed. (Cambridge University Press, 2014), http://www.mmds.org/, ch. 1.

Practicum: Using text2vec, create a document-term model of a corpus. Experiment with using words and n-grams, filtering terms and stemming, applying transformations such as TF-IDF, and applying distance measures such as cosine distance. Produce plots of how terms are used. (You may want to work ahead on unsupervised clustering or principal component analysis.)

Week 4 (October 5): Clustering and classification

• Lawrence M. Friedman, A History of American Law, 2nd ed. (New York: Simon & Schuster, 1985), pt. 2.
• Jockers, Text Analysis, chs. 11–12.
• Matthew L. Jockers and Ted Underwood, “Text-Mining the Humanities,” in A New Companion to the Digital Humanities, ed. Susan Schreibman, Ray Siemens, and John Unsworth (Wiley Blackwell, 2016), 291–306.
• D. Sculley and Bradley M. Pasanek, “Meaning and Mining: The Impact of Implicit Assumptions in Data Mining for the Humanities,” Literary and Linguistic Computing 23, no. 4 (2008): 409–424, http://llc.oxfordjournals.org/content/23/4/409.short.
• Leskovec, Rajaraman, and Ullman, Mining of Massive Datasets, ch. 7.
• Gareth James et al., An Introduction to Statistical Learning with Applications in R (Springer, 2013), chs. 2, 3, 10.

Practicum: Try a variety of unsupervised classification methods (e.g., K-means) and dimensionality reduction methods (e.g., PCA) if you did not do so last week. Using the caret package make a supervised classifier (using various machine learning methods) to predict some aspect of your text. You should use the corpus metadata for your classification labels.

Week 5 (October 19): Text reuse

• Friedman, A History of American Law, pt. 3.
• Ryan Cordell, “Reprinting, Circulation, and the Network Author in Antebellum Newspapers,” American Literary History 27, no. 3 (September 1, 2015): 417–445, doi:10.1093/alh/ajv028.
• David A. Smith, Ryan Cordell, and Abby Mullen, “Computational Methods for Uncovering Reprinted Texts in Antebellum Newspapers,” American Literary History 27, no. 3 (September 1, 2015): E1–E15, doi:10.1093/alh/ajv029.
• David A. Smith et al., “Detecting and Modeling Local Text Reuse,” in Proceedings of the 14th ACM/IEEE-CS Joint Conference on Digital Libraries (IEEE Press, 2014), 183–192, http://dl.acm.org/citation.cfm?id=2740800.
• Leskovec, Rajaraman, and Ullman, Mining of Massive Datasets, ch. 3.
• Kellen Funk and Lincoln Mullen, “The Spine of American Law: Digital Text Analysis and U.S. Legal Practice.” See also our working paper and the notebooks for the project.

Practicum: Using either the textreuse or LSHR packages, look for reused passages in the corpus you are working with. Create clusters or networks of those reuses.

Week 6 (November 2): Topic modeling

• Michael H. Hoeflich, Legal Publishing in Antebellum America (New York: Cambridge University Press, 2010), chs. 2, 3, 7.
• Journal of Digital Humanities 2, no. 1 (winter 2012): http://journalofdigitalhumanities.org/2-1/.
• Robert K. Nelson and Digital Scholarship Lab, University of Richmond, “Mining the Dispatch,” 2011, http://dsl.richmond.edu/dispatch/.
• David J. Newman and Sharon Block, “Probabilistic Topic Decomposition of an Eighteenth-Century American Newspaper,” Journal of the American Society for Information Science and Technology 57, no. 6 (2006): 753–767, http://onlinelibrary.wiley.com/doi/10.1002/asi.20342/full.
• Graham, Milligan, and Weingart, Exploring Big Historical Data, ch. 4.
• Jockers, Text Analysis, ch. 13.

Practicum: Using the textmineR, create topic models of the corpus that you are working with. Make plots of changes in topics over time.

Week 7 (November 16): Word-embedded models

• Barbara Young Welke, Law and the Borders of Belonging in the Long Nineteenth Century United States (Cambridge University Press, 2010), 1–94.
• Benjamin Schmidt, “Vector Space Models for the Digital Humanities,” October 25, 2015, http://bookworm.benschmidt.org/posts/2015-10-25-Word-Embeddings.html.
• Benjamin Schmidt, “Rejecting the Gender Binary: A Vector-Space Operation,” October 30, 2015, http://bookworm.benschmidt.org/posts/2015-10-30-rejecting-the-gender-binary.html.
• Michael A. Gavin, “The Arithmetic of Concepts: A Response to Peter de Bolla. Modeling Literary History,” September 18, 2015, http://modelingliteraryhistory.org/2015/09/18/the-arithmetic-of-concepts-a-response-to-peter-de-bolla/.
• T. Mikolov and J. Dean, “Distributed Representations of Words and Phrases and Their Compositionality,” Advances in Neural Information Processing Systems (2013), https://papers.nips.cc/paper/5021-distributed-representations-of-words-and-phrases-and-their-compositionality.pdf.
• Jeffrey Pennington, Richard Socher, and Christopher D. Manning, “Glove: Global Vectors for Word Representation.” in EMNLP, vol. 14, 2014, 1532–43, http://nlp.stanford.edu/pubs/glove.pdf.
• Matt J. Kusner et al., “From Word Embeddings to Document Distances,” in Proceedings of the 32nd International Conference on Machine Learning (ICML 2015), 2015, 957–966, http://www.jmlr.org/proceedings/papers/v37/kusnerb15.pdf.
• GloVe vignette from text2vec and wordVectors documentation.

Practicum: Train a word-embedded model on your corpus, or train several models on chronological or thematic subsets of your corpus. Use the model(s) to find distinctive use of language. More ambitious: use some of the more cutting-edge algorithms that use word vectors as inputs, for example, for finding document distances.

Week 8 (November 30): Named-entity recognition

• Welke, Law and the Borders of Belonging, 95–158.
• Cameron Blevins, “Space, Nation, and the Triumph of Region: A View of the World from Houston,” Journal of American History 101, no. 1 (June 1, 2014): 122–147, doi:10.1093/jahist/jau184.
• Benjamin Schmidt and Mitch Fraas, “The Language of the State of the Union,” The Atlantic (January 18, 2015), http://www.theatlantic.com/politics/archive/2015/01/the-language-of-the-state-of-the-union/384575/.
• Mitch Fraas and Benjamin Schmidt, “Mapping the State of the Union,” The Atlantic (January 18, 2015), http://www.theatlantic.com/politics/archive/2015/01/mapping-the-state-of-the-union/384576/.

Practicum: Using the openNLP or coreNLP packages, run named-entity recognition on your corpus. Plot the entities over time or space.

Assignments and expectations

You are encouraged to work with the Making of Modern Law: Legal Treatises corpus, which the library will provide. Alternatives include Chronicling America, 19th Century U.S. Newspapers, and the WPA former slave narratives. You may substitute another corpus after consulting with me.

Come prepared to discuss the readings assigned for each meeting. There are several kinds of reading. Works on legal history set the stage for the corpus we are investigating, so that you know what questions are worth asking. If you decide to work on a corpus other than MOML, plan on substituting your own readings for works in that category. Works on other historical subjects demonstrate how the methods we are studying can be applied to historical or literary historical questions. The methodological pieces, well, explain how the methods work. Don’t get hung up on the details of implementation: read to understand what the transformation is that the method accomplishes. Some of the methodological pieces are more practical, while others provide the theoretical basis. There is much more to read than I can assign, but the references below (also in a Zotero group library) contain many additional works.

Before each meeting, you will create an R Markdown notebook which—in prose, code, and figures—explores your chosen corpus of texts using the specified methodology. Get as far as you can with each method. You are always welcome to share code and ideas with other people in the class, though each person must turn in his or her own notebook. Submit all of the notebooks for the semester in a single GitHub repository. Name each notebook something sensible, like 08-named-entity-recognition.nb.html.

At the end of the semester, you will write a conference paper which makes a historical argument on the basis of the corpus you have chosen. Submit these papers as a separate GitHub repository. Prepare a one-page proposal by week 4, share a one-page statement of progress and problems by week 7, and submit the draft by 5 p.m. on December 16.

Grades will be assigned with 50% of the weight given to completing the readings and notebooks, and 50% to the final paper.

Acknowledgments and fine print

I have benefited from syllabi by Ted Underwood, Andrew Goldstone, and Ben Schmidt. Kellen Funk provided advice about legal history readings.

See the George Mason University catalog for general policies, as well as the university statement on diversity. You are expected to know and follow George Mason’s policies on academic integrity and the honor code. If you are a student with a disability and you need academic accommodations, please see me and contact the Office of Disability Services at 703-993-2474 or through their website. All academic accommodations must be arranged through that office.

References