Starting in 2021 and as part of an internship, I collaborated with Dutch type foundry Typotheque on an ongoing research project to inform their work for the Devanagari script. A survey was created to evaluate different graphic variants of Devanagari characters. Hundreds of people were interviewed, gathering data and recommendations not only on the recognizability of each form of the Devanagari characters, but also on the preferences of readers and users.

As a type foundry working with Devanagari, Typotheque wanted to learn about the contemporary use of the script

Colonial British authority was established in Calcutta, where the first printed types were created, coining the term Calcutta-style of Devanagari, also known as northern forms of Devanagari. Bombay, a major city on India’s west coast, rose to global prominence in the 1800s through maritime trade. Lying in the Marathi linguistic zone (i.e., text is written in Devanagari), Bombay gave birth to the Bombay-style Devanagari type.

Today, digital fonts copy traditions that were established in the metal type era, and it is not clear which of the Calcutta-style and/or the Bombay-style Devanagari are still preferred in their respective areas of use. While the history of the use of these forms has been documented, Typotheque wanted to gain an insight into current demographic and regional preferences of various forms of Devanagari.

The result was a survey of Devanagari readers, to provide an understanding of users’ preferences today, aimed to help design fonts better suited to speakers of Hindi, Marathi and Nepali.

For the survey, fourteen graphic variants that change most significantly according to region were selected, as documented in the separate historical research that was being conducted simultaneously by Karthik Malli: Devanagari – The Makings of a National Character.

Recognition, preference and education for each graphic variant were compared against other parameters like age, place of education, mother tongue and languages that the participant could speak.

Participants meta data was compared against their recognition, preference and education responses to find patterns. Pre-analysis was performed using Microsoft Excel and R. Open-sourced software JASP was used to perform all statistical studies.

The data showed conclusive results, with observable preference and recognition patterns across participants. To visualise these results, a series of graphs were designed. Together with colleague Liang Hai, the trends observed were plotted on to maps, bar graphs and piecharts.

All findings were published on Typotheque’s website, along with a thorough essay detailing the methods of this research.

The final outcome of this project was a recommendation sheet, in which we provide design suggestions for the three main target populations: Speakers of Hindi, speakers of Marathi and speakers of Nepali. We came up with a two-tier system for the suggestions: Recommended and acceptable. Additionally, the sheet includes insightful information about the findings, such as the effect of the reader’s age or region.