This week marks the beginning of a six week “choose-your-own-adventure” unit for our Digital Methods course. Instead of meeting over zoom each week, we have been assigned a list of topics to review to get acquainted with the types of methodology used in Digital Humanities. These topics include: Content Management, Text Analysis I, Text Analysis II, Mapping, and Visualization. The nature of this unit allows each student to learn the topics in whichever order they choose.
I am extremely interested in Text Analysis and it is one of the reasons I eagerly signed up for this course last semester – as this class is actually my first class outside of anthropology (apart from one required anatomy course). I am still relatively new to Text Analysis – I self taught myself some bits and pieces last year – but the experience really taught me what I don’t know and how much there still is to learn. I’m really excited to dive into the material this unit and will likely start with Text Analysis so that I can better frame new research projects that I can tackle in the future.
But, as it’s the beginning of our methodology unit, I thought I would use my blog today to lay out a project I have already done with Text Analysis so I can look back on it at the end of this semester to see how my understanding of it may have changed and advanced. Plus, this project is one that I will be presenting at a (virtual) conference this week for the American Academy of Forensic Sciences (AAFS). As this conference is virtual, I had to sign away my intellectual property on this project and any associated images so I would like to acknowledge AAFS’ ownership of this work before I proceed.
For this project, I wanted to examine the presence, impact, and structure of forensic anthropology articles within the American Journal of Physical Anthropology (AJPA). This journal is often regarded as a choice publication venue for those in forensic anthropology, as it’s impact factor is higher than those of Forensic Anthropology (FA) or Journal of Forensic Science. However, it may be questionable as to whether this journal is appropriate for forensic anthropological literature, as it’s geared towards an overarching biological anthropology audience (forensic anthropology is often considered a subfield of biological anthropology). By looking more closely at the forensic anthropology articles in the AJPA, we might be able to tell how these articles fit into the journal overall versus how they fit in with articles geared specifically to the subfield of forensic anthropology.
The sample for this project includes all articles within the AJPA that had readily available abstracts and all articles published in FA. For this project, I used R (and R Studio) to web-scrape article data (article titles, authors, dates, abstracts, keywords, etc.) using the “rvest” package. Next, all abstracts underwent Text Analysis via the “tidytext” package to assess word frequencies, correlations, and predictors (via term frequency-inverse document frequency (ti-idf) scores). Forensic anthropology articles in the AJPA were identified from their keywords and articles from FA were used as a neutral control during comparison of forensic and non-forensic AJPA articles.
Overall, I found that forensic anthropological articles make up less than 2% of all AJPA articles, but still earn high citation counts, as they represent 13% of the most-cited AJPA articles. In fact, they make up the top three most-cited AJPA articles overall.
Looking at word frequencies, forensic AJPA articles seem to better match FA articles than non-forensic AJPA due to their focus on individuals over populations and skeletal remains over humans in general. tf-idf scores were used to identify selectors for each category and of the top ten selectors in the non-forensic AJPA articles, four relate to primates and three to dentition. In contrast, the FA article selectors follow a strong trauma focus, with “fractography,” “kerf,” and “saws,” among the top ten. Forensic AJPA article selectors are heavily methodologically focused, featuring words such as “ages,” “pronasale,” “race,” “fusion,” and “indicator.” This seems to indicate that forensic AJPA articles align with the overall forensic anthropological focus on methodology and validation, but are more applicable to biological anthropology as a whole.
This project seems to indicate that forensic anthropology has a minimal but impactful presence within the AJPA, as the majority of articles accepted focus on methodology applicable across biological anthropology rather than just to the discipline of forensic anthropology. Of course, there is no significance test or p-value that can be used to truly determine such results, but I believe this type of project can open up discussions within the field regarding publishing forensic anthropological literature and what audiences we are really writing to per each journal.
And that’s a wrap! Thank you for reading my blog this week! If you’re interested, check out my poster here. And I would like to acknowledge the following resources that guided my project process:
- Bradley A, James RJE. 2019. Web scraping using R. Association Psych Sci 2(3):264-270. https://doi.org/10.1177/2515245919859535
- Silge J, Robinson D. 2020. Text mining with R: A tidy approach. Accessed at: https://www.tidytextmining.com/
