After a myriad of “R Session Aborted” messages, my soul might have crashed a bit but it did not die. It’s time to bring this R journey to the finishing line. Yet, every finish line is just a starting line in disguise.
This final blog post will serve as a larger assessment of the course and will address a fundamental question that arouse: “Are the text mining methods discussed (topic modeling, sentimental analysis, and word embedding) relevant, or are they mostly just superficial statistics detached from reality?”
Nan Da (2019) has suggested a critical analysis of computational literary studies (CLS). She criticized the use of computer programs in literary writings simply counting words and it’s all about frequency. She argues each text-mining research publication can only make simple arguments and frequently fails to do it appropriately, either by distorting evidence or failing to provide any evidence at all.
Da’s work drew several responses and criticisms. Her argument, “computational textual analysis has a threshold of optimal utility, and literature is that cut-off point,” (Da 2019, 639) stirred heated debates. One critique argues that Da fails to understand the statistical method as a contextual, historical, and interpretive project (Long and So 2019). Focusing on numbers is losing the context which is critical when interpreting literary sources. Another critique voiced that Da followed selective reading from the field of literary criticism (Piper 2020). According to this criticism, Da deliberately frames all computational approaches inside a single, restricted notion of significance testing, which she does not adhere to.
In chess, once thought to be a boundary for human intelligence, computers beat humans. This makes us wonder whether the computer will reach that chess moment in literary studies. Will the computational research methods catch up enough to analyze the literature? I argue that we should not dismiss CLS so quickly. Generator AIs such as painting drawings and producing podcasts demonstrate machine learning has gotten better over time and will improve further. I don’t agree with Da’s reference to CSL limitation as a ‘cut-off point’ as it limits the technology’s future application in academic research. Instead, we should think of current challenges as future research perspectives. For instance, if CSL is decentralized from a Western-centric database, it will become a more valuable research tool to analyze histories around the globe by grasping cultural context and hidden meanings. While discussing the possibility that computers are becoming better than humans at reading text, we should consider the ethical consequences. Where does intelligence reside and do we want computers to interpret the literature as human brains and minds do?
Preparing for my final paper and upcoming thesis research, I am nowhere close to writing code on my own in the R studio to draw valuable analysis yet. However, the course has taught me the valuable ‘R mindset,’ which consists of stealing a professional’s code, troubleshooting, acknowledging limitations, and not panicking when something fails. So that’s a wrap for my R programming journey (for now)!
Reference
Da, Nan Z. “The computational case against computational literary studies.” Critical inquiry 45, no. 3 (2019): 601-639.
Long, Hoyt and So, Richard Jean. 2019. “Computational Literary Studies: A Critical Inquiry Online Forum.” Accessed October 28, 2022. https://critinq.wordpress.com/2019/03/31/computational-literary-studies-a-critical-inquiry-online-forum/.
Piper, Andrew. “Do we know what we are doing?.” Journal of Cultural Analytics 5, no. 1 (2020): 11826.
Digital Tools for the Future 
Leave a comment