A whale of a tool: navigating texts with Voyant Tools
As a corpus linguist, I happen to teach my students how to use R for text analysis. However, due to the steep learning curve associated with R, I sometimes need to provide more accessible, ready-made alternatives. Two such tools I recommend are Lawrence Anthony’s AntConc and Stéfan Sinclair and Geoffrey Rockwell’s Voyant Tools. This post focuses on Voyant Tools, an open-source, web-based application for text analysis, ideal for both beginners and experienced researchers.
Voyant Tools in a few words
Voyant Tools is a web-based text reading and analysis environment designed to facilitate macro-reading and interpretive practices for digital humanities students and scholars. Its user-friendly interface makes it a perfect backup solution when time constraints prevent in-depth R training.
Getting Started with Voyant Tools
To begin using Voyant Tools, follow these steps:
Visit the Voyant Tools website
By default, Voyant Tools is in French: https://voyant-tools.org/. To access the English version: https://voyant-tools.org/?lang=en.

Input your text
Voyant Tools comes with pre-loaded corpora; William Shakespeare’s plays, Jane Austen’s novels, and Mary Shelley’s Frankenstein. You can access them by clicking ‘Open’, right below the ‘Add texts’ box.

Of course, the most interesting feature is to input your own text(s). You can either paste the text directly, preferably if it is not too long, enter a URL, or upload files from your computer. Supported file formats include plain text, HTML, XML, PDF, RTF, and MS Word documents. Here, I am using Herman Melville’s Moby Dick in .txt format, which can be downloaded from this link (the text is downloaded from Project Gutenberg; I have post-processed it with R).
Once your text is loaded, Voyant presents a multi-panel interface with several interactive tools:

Because Voyant Tools allows sharing its working environment, you should be able to play with the interface by following this link.
Key tools
Cirrus
Cirrus is a word cloud visualization showing the most frequent terms in your corpus. Right now, it does not tell us much because (a) there are not many words on the plot, and (b) most of the words are grammatical and do not tell us much about what the novel is about.

This can be fixed easily by hovering your mouse over the upper-right corner and clicking on the switch in the menu that appears (each tool has this option menu, and each menu is accessed this way):

This allows you to activate a stopword list, which will exclude common words that may clutter your analysis. Using the drop-down menu, make sure you select the language that matches the language of your text.

Here, we select English and click ‘Confirm’ to reload the word-cloud. The cloud becomes more meaningful.

Below, we set the number of words to 150 using the ‘Terms’ ruler.

Yes, Moby Dick is about a whale, and Captain Ahab is not lurking far.
Reader
Reader displays the full text of your documents for close reading and includes a search feature to help you find and examine specific terms in context.

The tool supports regular expressions, allowing you to fine-tune your searches.

Trends
Trends displays the distribution of terms across your corpus or document segments. The figure below compares the distributions of five terms (whale, sea, old, man, and like) across the corpus, which is divided into ten sections.

Options allow the user to select one specific term (like Ahab below) and choose from a variety of graph types (Area, Columns, Line, Stacked Bar, or the default Line + Stacked Bar).

What I like about this novel is that, except for segment 3 where Ahab and the whale co-occur, segments 5 to 8 show whale as quite frequent and Ahab as rare. However, from segment 8 onwards, Ahab peaks while whale does not. This nicely suggests the back-and-forth between Captain Ahab and the whale.
Summary
As its name indicates, summary provides an overview of your corpus, including word count and distinctive terms. Note that the statistics are influenced by the use of a stopword list. The Documents tag is useful if your corpus consists of several distinct parts. The Phrases tag allows the user to spot recurring multiword expressions in the corpus.

Contexts
Context is familiar to you if you are used to KWIC tools. It displays keywords in context (below, the keyword is whaler), allowing you to examine how specific terms are used throughout the text. Again, it comes with a bunch of options. You can therefore expand or reduce the context.

Collocates
The Collocates tag allows you to examine words that frequently appear near the keywords in your corpus.

The context ruler is used to expand or reduce the search context for your collocates. You can also select a specific keyword and see its most distinctive collocates.
Additional tools
Clicking on the windows logo at the top-right corner of the Voyant Tools interface will give you access to more expert tools.

Navigate the drop-down menu to reveal the list of available tools.

I describe two of them briefly, but there are many more. A list of tools is available here. Feel free to explore! My own experience tells me that some tools are more like gadgets than real aids to data exploration. My recommendation is therefore to use a tool not because its graphic output looks nice, but because it reveals some important aspect of your data in an efficient way.
ScatterPlot
My favorite tool, by far because it relies on exploratory techniques that I use in my own research, namely principal component analysis, correspondence analysis, and t-SNE.1 To access it: windows logo > Visualization tools > ScatterPlot.
ScatterPlot helps visualize statistically significant associations between terms in your corpus/corpora based on their distributions.
The plot below relies on correspondence analysis. The size of each data point (=each word) depends on its frequency. The position of each data point depends on its distribution in the part of the corpus considered. Here, the corpus is split into ten “bins” (=ten parts).

More on ScatterPlot here.
Collocates Graph
Collocates Graph visualizes keywords and terms that frequently appear together as a force-directed network graph. More on network graphs in these two posts: here and here.
The Context slider controls the number of terms included when searching for collocates. The selected value represents the number of words considered on each side of the keyword, effectively doubling the window of words. By default, the context is set to 5 words per side, with a maximum of 30. In the example below, I have set the context to 13 words and chosen ‘whale’ as the keyword.

Assets and Limitations
Voyant Tools offers a powerful, accessible platform for text analysis, making it an invaluable resource for both beginners and experienced researchers in corpus linguistics. By integrating Voyant into your toolkit, you can efficiently explore and analyze textual data, uncovering patterns and insights that might not be immediately apparent through manual analysis.
Voyant makes it easy to share your findings. You can export visualizations, data, or entire tool configurations. This feature is particularly useful for incorporating results into publications, presentations, or further analysis.
While Voyant excels at quantitative analysis and distant reading, it does not replace the need for close reading or qualitative interpretation. Researchers should use Voyant in conjunction with traditional methods for a comprehensive understanding of their texts.
Video recap
If you have one hour to spare, I highly recommend that you take a look at this step-by-step video guide published by the University Libraries of the University at Albany.
References
- https://voyant-tools.org/docs/
- https://infoguides.gmu.edu/textanalysistools/voyant
- https://brockdsl.github.io/Voyant-Tutorial/
- https://guides.library.ucsc.edu/DS/Resources/Voyant
- https://www.youtube.com/watch?v=4jCGLmbLFT0
- Another method, Document Similarity, is proposed, but it is really useful when your corpus consists of several documents. [↩]
OpenEdition vous propose de citer ce billet de la manière suivante :
Guillaume Desagulier (9 août 2024). A whale of a tool: navigating texts with Voyant Tools. Around the word. Consulté le 18 juin 2025 à l’adresse https://doi.org/10.58079/125pn