This classroom-based activity will introduce students to models trained on corpora from different centuries using the Word Vector Interface (WVI)—available through the Women Writers Vector Toolkit (WWVT)—and enable them to begin exploring larger thematic questions about the role of chronology in Women Writers Online (WWO). The activity can serve as a starting point for incorporating WWO, chronological analysis and comparison, word embedding models, and text analysis in the classroom.
This activity will provide an opportunity for learners to explore WWO and perform comparative analyses on collections of texts by women writers in the 16th+17th, 18th, and 19th centuries. This type of activity will be useful in courses that focus on early women writers or early literate culture more broadly.
This activity is broken down into several parts. First, students use the “Cluster” function to look at clusters of related terms in models trained on texts from a particular century. Next, the students begin querying terms of interest from the model. If there is time, the students will then conduct a comparative analysis by looking at results for a model from a different century. The scope of this activity varies based on the time available; it is recommended to allocate at least an hour for meaningful exploration, analysis, and discussion.
This activity will enable students to:
- Gain a basic understanding of the Women Writers Word Vector Interface, particularly the “Clustering” and “Basic” query options.
- Examine how particular words are used in the 16, 17th, 18th, and 19th centuries.
- Conduct a comparative analysis of two centuries’ models.
- Explore how the relationship between words in a corpus tie to larger cultural values of the time.
Activity 1: Clustering by Century
For this activity, your students will explore the “Clusters” results for the model of choice (WWO 16+17th, WWO 18th, or WWO 19th). The “Clusters” results show 10 randomized clusters from your model. Every time you click “Reset Clusters,” you get a new randomized set of clusters. For more information about “Clusters” results and “Basic” query results, we encourage you to check out this activity.
Clusters provide a way for students to quickly discover sets of related terms within a model, and begin to think about the how word embedding models place words within vector space.
To get started, ask students to select one of the chronological models in the left column of the interface and then click “Clusters.” Have students search through the clusters, continuing to refresh them until they find a particularly interesting one. You might recommend that they save these clusters for future exploration and analysis.
Once your students have identified at least one cluster of interest, encourage them to use the “Basic” query function to explore words from that cluster. Ask students to have a small-group or full-class discussion about the results of the “Basic” query function for the particular words they’re interested in, focusing on results that are surprising or unexpected. During these conversations, encourage students to think about what these general results may say about the texts from the century they are exploring.
Finally, ask students to click on words of interest, to see how they are used within the WWO corpus and compare that usage with their expectations from exploring with the Word Vector Interface. Note that students can narrow their results in WWO to the century they are exploring with the timeline or century-selection checkboxes.
Activity 2: Century Comparison Using Clustering
Ask students to select a different century from the chronological models and repeat their searches with the “Basic” query option. Ask them to discuss how the related terms they discover are the same or different for models from different centuries. Discuss particularly surprising or interesting differences as a group, and ask students to consider what these might show about the cultures and time periods reflected in the texts.