Mapping Armenian women’s voices: A data-driven look at their writing
A few months ago, I interviewed Dr. Victoria Rowe about her dissertation on Armenian women’s writing in the late 19th and early 20th centuries. Our conversation, featured in the February 2025 issue of Torontohye, left me deeply impressed by her research and curious. Was there any recent analysis of modern Armenian women’s writing?
Google Scholar didn’t offer many answers, so I decided to take matters into my own hands and analyze Armenian women’s writing myself.
Data Processing
Scope of analysis
To keep my research focused and manageable, I narrowed the scope of my analysis to my immediate counterparts: Armenian women who write in English for Armenian news publications. I also limited my dataset to articles that were easily accessible online and published in recent years. To gather this data, I fully scraped three websites— EVN Report, the The Armenian Weekly, and the The Armenian Mirror-Spectator—and partially scraped Horizon Weekly. The first graphic (FIGURE 1) illustrates the number of articles collected from each source, the proportion of pieces credited to individual contributors, and the likely gender of those contributors.
Limitations of data collected
For each article scraped, I recorded key details: the published date, URL, author’s name, and also keywords. The keywords were generated using a simple form of natural language processing (NLP)—a method that enables computers to process and interpret human language.
Since I was working with a fairly large dataset without specialized hardware, my analysis remained surface-level. Instead of diving into the nuances of tone or perspective, I focused solely on identifying common keywords in each article. For example, I could track how often “EU” appeared as a major keyword, but not whether Armenian women writers were praising or critiquing the EU’s relationship with Armenia.
Limitations of gendering the dataset
Once I had standardized data and extracted the authors' first names, I needed to determine the likely gender of each writer to focus specifically on Armenian women’s contributions.
This process came with some inherent challenges. First, I assumed a binary gender classification (‘male’ or ‘female’), which meant I wasn’t accounting for possible non-binary contributors. Second, I assumed an author’s gender aligned with the most commonly associated gender of their first name—a method that doesn’t account for gender-neutral names (including gender-neutral names like ‘Sam’ and using first initials like ‘C. S. Lewis’). Finally, my analysis only included articles where a named individual was credited as the author. This means that pieces written by women as part of editorial teams, collective statements, or anonymous contributions were not included in this analysis.
I hoped that my dataset was large enough that occasional anomalies would not significantly skew the results.
When it came to gendering the dataset, I couldn’t rely on standard services for large-scale name classification. One reason was cost—many of these services charge fees that add up quickly. But more importantly, most of them struggle with gendering Armenian names. This is likely due to the skewed dataset they use—one popular service has only 2,560 Armenian names in its dataset, compared to hundreds of thousands for anglophone countries (Canada: 26,187, United Kingdom: 105,368, United States: 169,208). Any AI-powered service needs to be trained on lots of data to be accurate, and a biased sampling of data can lead to inaccuracies.
I found that OpenAI’s GPT chat models were able to accurately gender both Anglophone names and Armenian names, so that’s what I settled on using. Gendering thousands of author names ended up costing less than a cup of Tim Hortons coffee.
Biases and limitations when categorizing keywords
To get a quick visual overview of the major keywords and themes in Armenian women’s writing, I put together a word cloud (pictured below). However, I was not satisfied with just seeing which words appeared most often—I wanted a deeper understanding of how different topics compared in prominence.
To do this, I categorized the most frequently occurring keywords into broader themes. This allowed me to analyze what Armenian women were writing about when they covered geographical locations, socio-political issues, and cultural matters. Sorting the keywords into categories helped reveal patterns, but it also introduced a level of subjectivity, as I had to decide which words fit where.
The method I used for categorizing keywords was inevitably shaped by my own biases. I created a list of categories based on patterns I noticed in the most frequently used keywords (countries, cities, blockade and conflict, political terms, economic terms, etc). I chose ‘seed’ words for each category and set a similarity threshold to determine which terms would be grouped together.
While I did iterate and experiment with different choices for categories, seed words, and similarity thresholds until I was satisfied with the accuracy of the result, since there was no set standard I followed, there is inevitably room for improvement.
Data Takeaways
Analyzing the data revealed several key insights into the ‘typical’ style of writing featured in Armenian publications.
- Armenian publications often don’t cite individual contributors
One of the most surprising findings was the lack of bylines in many articles. More than half of the pieces I scraped did not credit an individual author. Instead, they were attributed to the editorial team, an external organization (such as ARS and ANCA) or anonymous ‘guest contributors.’ This significant lack of authorial accountability and transparency was unexpected.
Of course, there’s nothing inherently wrong with publishing press releases, collaborative reports, or anonymous pieces—they can be valuable additions that bring diverse perspectives to a newspaper. However, when these types of contributions dominate original content, they pose a real risk to journalistic diversity, critical coverage, and overall quality of publications.
If Armenian publications are largely recycling the same press releases and republishing articles instead of offering fresh analysis or local reporting, what’s the point of duplicating work and using up resources?
-Armenian newspaper writers are likely disproportionately men
This finding, while not surprising, is still disappointing. Among the articles I scraped with individual contributors, less than a third were written by likely female authors.
There are a few possible explanations. Perhaps Armenian women are more likely to contribute without individual credit—as part of editorial teams, organizations, or under anonymous bylines. It’s also possible that rather than write about current events for Armenian publications, Armenian women writers are contributing their skills elsewhere. While I can’t say for certain that there are more male than female Armenian writers overall, the data suggests that men are disproportionately represented in bylined contributions.
It’s still important to have Armenian women writing for Armenian publications as they can help highlight unique gender issues from the perspective of someone directly impacted.
- Armenian women mainly focus on Armenians (and their neighbours)
The data revealed a strong geographic focus in Armenian women’s writing. The most common geography keywords were ‘Armenia’ and ‘Artsakh,’ and the most common nationality keyword was ‘Armenian.’ The most common country keywords were of countries surrounding Armenia (‘Azerbaijan,’ ‘Russia,’ etc). I could oversimplify and accuse Armenian women writers of being ethnocentric, but I think this takeaway is a result of the data sources I used. I specifically analyzed Armenian women writing for Armenian publications, where the topics are naturally centered on the current events impacting the Armenian nation. Armenian women who write personal blogs, fiction, or other non-political content were largely excluded from the dataset.
Still, this finding is significant—it highlights that Armenian women writing for publications are interested in national and regional affairs impacting the modern Armenian state.
- Armenian women are influenced by world events
Tracking keyword trends over time revealed that Armenian women’s writing is responsive to global events. For instance, I noticed ‘Lebanon’ spiked as a keyword during the year of the Beirut Blast, while mentions of ‘Ukraine’ increased after Russia’s 2022 invasion. This correlation points to Armenian women’s reporting potentially being influenced by world events.
Armenian women focus on Christianity when writing about religion
I had seeded the religion keywords category with words from all major religions. However, the overwhelming majority of keywords in this category were related to Christianity. One of the most frequently mentioned words was ‘church,’ which isn’t surprising given the central role churches play in Armenian community life.
Interestingly, the keywords also highlighted diversity in the type of Christian denominations featured. Armenian women write about Evangelical, Protestant, Catholic, Orthodox, and Apostolic religious traditions. This suggests that while Christianity remains a dominant theme, the coverage acknowledges and explores its different expressions within the Armenian world.
Future Work (and what it would require)
There are several ways to expand on this research and develop a more comprehensive analysis of Armenian writing.
Larger dataset (i.e., scraping more websites)
Expanding the dataset would allow for stronger conclusions about broader trends in Armenian women’s writing. This would involve scraping more websites (which would require creating unique web scrapers for each one). Fortunately, my existing data processing pipelines are scalable to be able to integrate new data from websites.
- Analyzing Armenian women’s writing in Armenian, Russian, French, and other languages
Since my study was focused on English-language Armenian writing, it naturally excluded a large community of Armenian women who write in Armenian, Russian, French, and other languages. Expanding our analysis to the global community of Armenian women writers would be a natural next step.
To do this, we would need to scrape the websites of non-English publications. We would also need NLP methods capable of extracting and categorizing keywords in non-Latin languages like Armenian and Russian.
- Deeper analysis of article content
I lacked the skill to do a deep dive into the tone and perspective taken on topics. I would love to collaborate with someone more skilled in advanced NLP techniques to write scripts capable of detecting sentiment, tone, and perspective within the articles.
For example, while my analysis showed that ‘church’ is a common topic, it couldn’t determine whether Armenian women were writing positively about community experiences, critiquing certain traditions, or engaging with religious themes in other ways. Automating this type of deeper analysis would help reveal broader patterns in how key topics are approached.
What can we do with what we know so far?
While my analysis is just a starting point, we do have some key takeaways that allow us to take action to improve the quality and diversity of writing in Armenian publications.
1. Increase authorial accountability
Publications should make a greater effort to credit individual authors in bylines. This would help build trust with readership, give credit to the writers who work hard to produce original content, and allow us to more easily gauge the level of diversity in newspaper contributors. Readers can also play a role—if they notice their favorite publication regularly omitting author names, they should reach out to the editorial team and express their concerns.
2. Encourage more Armenian women to write
To ensure that Armenian women’s voices are represented, financial and social support is crucial. Having grants and honorariums available to writers would remove some of the barriers to devoting time to writing. Specifically, soliciting women’s perspectives on issues, prioritizing gender parity on editorial teams, and having mentorship and guidance available would encourage women to contribute to their local Armenian papers.
3. Invest in the study of the modern
We should direct more resources towards analyzing modern Armenian cultural products. A lot of Armenian studies research seems to focus on medieval manuscripts and the 1915 Genocide; however, cultural studies could benefit from analyzing recent trends. The last 30 years of Armenian history have been packed with important milestones and events, each of which has inspired news coverage and cultural output. It is worthwhile to critically examine the cultural landscape these outputs have created and decide if the status quo best serves the Armenian community.
Where do we go from here?
Armenian women’s writing is an essential cultural product of modern Armenian society. The trends in media coverage reveal both strengths and risks, and it’s up to the broader Armenian community to continue researching these patterns and implementing meaningful changes. The creation of quality cultural publications—where women’s voices are equitably represented—is a shared responsibility of the collective community. ֎
This analysis was published in Torontohye's April 2025 (#212) issue.
