I’ve been working on an update to the tentative analysis I did on the black woman suffrage database to present at Women’s History in the Digital World at the end of this month. The database is part of the full text sources in Women and Social Movements. [disclosure: My essay is not behind a subscription log in. While the project is fee-based, the subscriptions are very low priced, and there are free trials. I am the digital history editor for the site and am actively soliciting digital projects using the full text sources.]
Working with twice as many files (now 904 items) as I had for the first project (more details on corpus), I realized that the embodied references I had noted as distinctive in the writings of black suffragists as opposed to the almost 100% white History of Woman Suffrage came in large part from the writings of Frances Ellen Watkins Harper. I therefore decided to separate her writing out into fiction and non-fiction and compare those sub corpora to the remaining files which I sorted into fiction and not-fiction. [54 items for Harper v 850 items in Black Woman Suffrage Database BWSD, by word count 280K v 2.1M)
In the original project I wondered if the embodied references were part of a genre convention, particularly of sentimental fiction so strongly associated with white women's writing in the 19th century. The function of sentimental fiction is to reduce the gap between subject and reader through empathy, bringing the reader to see where justice lies by foregrounding “moral purpose.” A key marker of sentimental fiction is emphasis on suffering according to Frances Smith Foster, while Barbara Christian argues that it is “an excessive display of emotion." Because semantic tagging can identify both social relationships and processes, as well as emotion, these two criteria will work as markers of sentiment.
Harper worked within the confines of a genre reliant on heavily gendered tropes to illustrate the ways black women should be included in, but are excluded from, ideals of true womanhood. In a sense, this sentimental discourse is the literary expression of what she so succinctly noted at the infamous 1869 meeting of the American Equal Rights Association "the white women all go for sex, letting race occupy a minor position." [I'm working largely from C.C. O'Brien brilliant analysis in "The White Women All Go for Sex": Frances Harper on Suffrage, Citizenship, and the Reconstruction South.]
In the original project I wondered if the embodied references were part of a genre convention, particularly of sentimental fiction so strongly associated with white women's writing in the 19th century. The function of sentimental fiction is to reduce the gap between subject and reader through empathy, bringing the reader to see where justice lies by foregrounding “moral purpose.” A key marker of sentimental fiction is emphasis on suffering according to Frances Smith Foster, while Barbara Christian argues that it is “an excessive display of emotion." Because semantic tagging can identify both social relationships and processes, as well as emotion, these two criteria will work as markers of sentiment.
Harper worked within the confines of a genre reliant on heavily gendered tropes to illustrate the ways black women should be included in, but are excluded from, ideals of true womanhood. In a sense, this sentimental discourse is the literary expression of what she so succinctly noted at the infamous 1869 meeting of the American Equal Rights Association "the white women all go for sex, letting race occupy a minor position." [I'm working largely from C.C. O'Brien brilliant analysis in "The White Women All Go for Sex": Frances Harper on Suffrage, Citizenship, and the Reconstruction South.]
As Hazel Carby notes, Harper crafted a discourse for black women that “address[ed] exclusion from the ideology of true womanhood” and “rescue[d] their bodies from a persistent association with illicit sexuality.” That she did this by relying in part on embodied references drawn from sentimental conventions makes perfect sense. Fortunately, semantic tagging is also quite good at locating content related to the body, giving me a third criteria to use in assessing sentimentality.
I wondered how Harper compared to other women writers of her era, both black and white, beyond the limited corpus I had initially used in the History of Women Suffrage that included very little fiction. Were her embodied references similar to other sentimental women's writing? Did the sentimental seep into her nonfiction as well as her fiction?
Sara Steger in Patterns of Sentimentality in Victorian Novels provided a great starting point and an inspiration. I ran the files through Antconc, a concordancing software, and Wmatrix, a semantic tagging program. Steger explored the most distinctive adjectives, nouns, and verbs in a sentimental versus a non sentimental corpus (which was British English not American English). Using wmatrix relative frequencies, I first looked at overlaps in the various fiction sub-corpora to address the question of Harper's similarities to other writers, and then selected the unique words from Harper's corpus to explore.
This analysis revealed two things. First using the baseline established by Steger, which is not a perfection comparison, but gave me something to work from, the black women's writings were largely distinctive save for nouns which had a high degree of overlap with the sentimental words identified by Steger. This isn't particularly surprising as my corpus consists of authors involved in suffrage and abolition, while Steger's corpus was simply fiction. One possible explanation for the high degree of overlap in the nouns may be that these are simply common nouns of the era. I need to check the Corpus of Historical American English fiction sub corpus for those numbers still. [update indeed these are highly frequent nouns although for individual authors more precise numbers need to be calculated as I do below].
Figure 1A-1C Most most frequent words by part of speech in Harper's Fiction, Fiction by all other authors in BWSD, and Steger's findings, which are most distinctive adjectives (by log likelihood) in sentimental NOT most frequent.
This analysis revealed two things. First using the baseline established by Steger, which is not a perfection comparison, but gave me something to work from, the black women's writings were largely distinctive save for nouns which had a high degree of overlap with the sentimental words identified by Steger. This isn't particularly surprising as my corpus consists of authors involved in suffrage and abolition, while Steger's corpus was simply fiction. One possible explanation for the high degree of overlap in the nouns may be that these are simply common nouns of the era. I need to check the Corpus of Historical American English fiction sub corpus for those numbers still. [update indeed these are highly frequent nouns although for individual authors more precise numbers need to be calculated as I do below].
The second thing this comparison illustrated was how similar Harper's fictional writing is to other Black women of her era with overlaps ranging for 40% (nouns) to 80% (adjectives).
Highlighted in yellow are the words that were unique to Harper's writings that seemed like they might be connected to a sentimental discourse. poor and hope invokes issues of justice, while face and hand refer to embodiment. [I still need to confirm log likelihood values for these words]
Once again I began by contexualizing her her nonfiction writing against nonfiction in the BWSD. Overlaps here ranged for 30% (verbs) to 60% (nouns). The unique results seemed consistent with what I'd already started to see in running comparisons in Antconc, specifically that there was a discourse of moral power in Harper that is tied to sentimental conventions. I noted as well the repeated appearance of hope, as well as appeal, ask, hear all seemingly connected to efforts to get the reader or audience involved with her discourse.
Combining the highlighted terms from machine driven inquiry in the fiction and the nonfiction corpora, I created a list of words that seemed connected to sentiment and were unique to Harper. I compared relative frequencies and log likelihood values, a statistical measurement of the significance of the observed differences. These will give me places to start with close reading. Below is an illustration of these words in Harper (lavender) fiction (solid) and nonfiction (patterned) and all the other BWSD (blue) fiction (solid) and nonfiction (patterned). Log Likelihoods are below the word on the X axis. They are a little tricky to read as they refer to the two variables with the greatest gap. In the case of power, where Harper and the other writers are quite similar in the nonfiction, the gap is observed between her nonfiction and BWSD fiction. Human and moral on the other hand are over-used by Harper in her nonfiction. Heart one of Harper's favorite words is overused in both her fiction and her nonfiction. [updat: Looking back at this I realized I needed to re-run these numbers for fiction and nonfiction separately. Updating that soon]
However, looking at individual words that refer to body parts shows that Harper used heart (LL 5.3) more than Stowe did. I also checked power, which again is used more by Harper (LL 9.4 ) This seems to suggest that within Harper's sentimental discourse of the novel is also this idea of moral power.
Once again I began by contexualizing her her nonfiction writing against nonfiction in the BWSD. Overlaps here ranged for 30% (verbs) to 60% (nouns). The unique results seemed consistent with what I'd already started to see in running comparisons in Antconc, specifically that there was a discourse of moral power in Harper that is tied to sentimental conventions. I noted as well the repeated appearance of hope, as well as appeal, ask, hear all seemingly connected to efforts to get the reader or audience involved with her discourse.
A final way of exploring this question is to compare Harper's fiction and non fiction writing. Overlaps here ranged from a high of 50% for adjectives to a low of 40% for verbs. It took a long time to make sense of this, reliant more on close reading and examination of sub corpora, but it turns out that important words here that run through both fiction and nonfiction include good, life, hope, know, all of which are connected to Harper's sentimental discourse of moral power. moral and power are more frequent in the nonfiction, but by keyness, which takes into account the size of the corpora, more to nonfiction, but as my exploration reveals below,
Combining the highlighted terms from machine driven inquiry in the fiction and the nonfiction corpora, I created a list of words that seemed connected to sentiment and were unique to Harper. I compared relative frequencies and log likelihood values, a statistical measurement of the significance of the observed differences. These will give me places to start with close reading. Below is an illustration of these words in Harper (lavender) fiction (solid) and nonfiction (patterned) and all the other BWSD (blue) fiction (solid) and nonfiction (patterned). Log Likelihoods are below the word on the X axis. They are a little tricky to read as they refer to the two variables with the greatest gap. In the case of power, where Harper and the other writers are quite similar in the nonfiction, the gap is observed between her nonfiction and BWSD fiction. Human and moral on the other hand are over-used by Harper in her nonfiction. Heart one of Harper's favorite words is overused in both her fiction and her nonfiction. [updat: Looking back at this I realized I needed to re-run these numbers for fiction and nonfiction separately. Updating that soon]
I also made use of the semantic tagging functions in Wmatrix, visualizing all the "E" tags, which have to do with emotion, to try to compare via distant reading the content of the works in various corpora. This is a preliminary view and what I still need to do is put these semtags in context for each of the corpora, but it is clear that in terms of evoking emotion, Harper is higher in all areas except happiness.
I wanted to get a bigger picture of Harper against white women's writing. Harper’s novel Iola Leroy(red) is often compared to Uncle Tom’s Cabin (blue), so I compared those two texts using semantic tagging in Wmatrix. I was particularly interested in this in light of Jockers claim that Stowe writes like a man despite her frequent association with the sentimental mode.
Looking at tags for social actions, States and Processes (S tags), Iola Leroy contains more content in 11 of the 13 tags that appear in the corpora. For Emotional Actions, States And Processes (E tags), the results are weaker, but Iola Leroy still contains more content in 4 of the 6 tags. Finally looking at tags for anatomy & health (B1, B2 tags) to get at embodied metaphors, the results are split evenly.
Looking at tags for social actions, States and Processes (S tags), Iola Leroy contains more content in 11 of the 13 tags that appear in the corpora. For Emotional Actions, States And Processes (E tags), the results are weaker, but Iola Leroy still contains more content in 4 of the 6 tags. Finally looking at tags for anatomy & health (B1, B2 tags) to get at embodied metaphors, the results are split evenly.
However, looking at individual words that refer to body parts shows that Harper used heart (LL 5.3) more than Stowe did. I also checked power, which again is used more by Harper (LL 9.4 ) This seems to suggest that within Harper's sentimental discourse of the novel is also this idea of moral power.
I feel somehow that these two things that the semtag clouds reveals, an embodied sentimental discourse of the fiction and a moral power discourse of the nonfiction were somehow two sides of the same coin.
I did a little more distant reading, looking at the 10 most frequent semantic tags in Harper's fiction and nonfiction. Excluding the grammatical bin and pronouns, which are so large they obscure everything in any visualizations I do, there is 60% overlap in the tags, indicating that in terms on content, Harper's writing is fairly similar.
I was particularly interested in the presence of the tag for body parts (anatomy and physiology) in both corpora since that was the issue that initially drew me into this investigation (although it is more present in the fiction corpus with a LL value of 32). Looking further into the words making up this tag, I found that hearts and hands, are present more in the nonfiction and in fiction, heart and hand predominate. That gave me the focus for my close reading.
I was particularly interested in the presence of the tag for body parts (anatomy and physiology) in both corpora since that was the issue that initially drew me into this investigation (although it is more present in the fiction corpus with a LL value of 32). Looking further into the words making up this tag, I found that hearts and hands, are present more in the nonfiction and in fiction, heart and hand predominate. That gave me the focus for my close reading.
No comments:
Post a Comment