






‚tÿpøf¡x": More Than Just a Typographic Proofreader
Typographic conventions governing quotation marks, numerals, punctuation, and spacing reflect the cultural heritage of individual languages. However, globalization and standardized software often lead to these conventions being overlooked. In an interview with Filip Blažek and Radek Sidun, we explore the origins of Typofix, a project that combines international research with technological development. The project has produced a database of typographic rules for all 24 official languages of the EU, a printed handbook, and an application designed to preserve the unique typographic conventions of each language. The project was jointly supported by UMPRUM and the Technology Agency of the Czech Republic through its SIGMA program.

Filip Blažek and Radek Sidun.
Photo: Apolena Typltová
What was the original inspiration behind the Typofix project? Which of you came up with the idea first, and why?
F
Filip Blažek
R
Radek Sidun
R
We sometimes joke that it was laziness. After all, who wants to go through a text correcting dozens of typographic errors one by one, while also taking into account the language it's written in?
F
In fact, the idea grew out of our experience with a similar tool that we had developed and used successfully for almost 20 years when typesetting Czech and English texts. We felt it would be a shame not to develop a more comprehensive version that could support a wider range of languages.
One of the premises of the project is that globalization and the dominance of English in international communication are weakening language-specific typographic conventions. What do you think is lost when texts in different languages begin to follow the same rules?
F
Whether we like it or not, the English-speaking world has a significant influence on us. We encounter it everywhere. Just try typing Czech quotation marks in Google Docs, for example. Ignoring local conventions breeds mistrust and indifference, along with a largely justified sense that many global companies display ignorance and an attitude of superiority towards the rest of the world.
Can linguistic and typographic differences also be considered part of our cultural heritage? Why is it important to preserve them?
F
Absolutely! Typographic conventions carry traces of history, as well as cultural and commercial ties. They are passed down from generation to generation, and the professional community has worked to preserve them even through difficult times.
‚tÿpøf¡x":
„Dónde están los baños , por favor?“
«¿Dónde están los baños, por favor?»
Typofix
Examples
Before/After
Spanish
She said—I hope—“my ‘birdie,’ I love you.”
She said – I hope – ‘my “birdie”, I love you’.
Typofix
Examples
Before/After
English
La caissière du cinéma m ’ a recommandé un “film sensationnel”!
La caissière du cinéma m’a recommandé un « film sensationnel » !
Typofix
Examples
Before/After
French
The aim of the project was to document the typographic conventions of all 24 official languages of the European Union. How did you approach the research, and who did you collaborate with? Were there already comprehensive resources you could draw on, or did you have to establish and standardize the rules for some languages for the first time?
R
The lack of accessible information was one of the key factors that led to the creation of Typofix. How am I supposed to typeset a Bulgarian or Greek text correctly when the typographic rules for these languages are either unavailable or only documented in the languages themselves? And even when they are available in English, they are often full of errors or incorrect symbols, or different sources contradict one another. AI isn't much help either: one model claims there should be a space between a number and the % symbol, while another insists there shouldn't be. We realized that the only way forward was to work with experienced local typographers, typesetters, and educators who could help us establish a consistent set of rules.
A number of international experts contributed to the project. Could you describe the research process, how these experts were involved in collecting the data, and how the findings were put into practice?
F
Initially, we thought the experts would populate our database of rules themselves. But the deeper we got into the complexities of regular expressions, the clearer it became that most of them would find this too challenging. So we simplified the data collection process, creating tables that described a wide range of typographic conventions and asking the experts to select or provide the correct variants for their respective languages. We then generated test files over several rounds, which the experts reviewed and commented on.

Workshop with international experts at UMPRUM.
Photo: Apolena Typltová
Did you encounter any surprising differences or similarities between languages during your research? Was there a particular language that surprised you? Could you give a specific example that illustrates why typographic rules cannot simply be transferred from one language to another?
F
We discovered many unexpected conventions. For example, some languages don't have the concept of nested quotation marks and simply repeat the same symbols for a quotation within a quotation. In others, a dash is used in much the same way as parentheses, meaning that a closing dash is required even if the phrase appears at the end of a sentence. We also found that in several languages, years are treated as ordinal numbers and therefore always followed by a period, although using a period to indicate ordinal numbers is relatively uncommon among European languages. Another surprising discovery was that no two languages we analyzed followed exactly the same typographic rules. There were always differences of some kind.
Language standards evolve, and some conventions allow for multiple approaches depending on the context or type of publication. How did you account for this variation in a database intended to serve as a reliable foundation for automated corrections?
R
We addressed the issue of alternative conventions from the very beginning, since we know from Czech that opinions differ on things like the appropriate length of a dash. In some languages, users can only choose between different styles of quotation marks, while in others, they can select different spacing widths, thousands separators in numbers, preferred date formats, or whether short prepositions and conjunctions should be joined to the following word.
F
The entire system is designed to be highly flexible. For example, if someone requests the option to use a different preferred pair of nested quotation marks in Danish, we can implement it in just two minutes.
One of the project's main achievements is translating specialist knowledge into a database that software can process. How did you convert linguistic and typographic conventions into machine-readable rules? What was the most challenging aspect of this collaboration between typographers, language experts, and developers?
R
The conversion process itself was quite demanding, as we had to make sure that correcting existing errors wouldn't introduce new ones into the text. Rather than creating separate rules for each language, we developed a system of variables. These might include month names, abbreviations of books of the Bible, or conventions for writing musical keys, which differ considerably from Czech conventions in many countries. Another type of variable covers incorrect variants, such as the various forms of apostrophes, multiplication signs, or degree symbols. Once we had mastered writing even highly complex expressions, ChatGPT proved helpful, having finally learned how to create and modify regular expressions in 2026. In the end, however, the hardest part of the entire process wasn't converting the collected data into rules, but getting the experts to respond to our emails…


Typofix has the potential to serve not only typographers and graphic designers, but also translators, proofreaders, editors, and everyday users. Who is the tool primarily intended for, and how did the needs of these different groups influence its development? What features will it offer at launch, and what makes it unique?
F
Initially, Typofix will probably be used mainly by graphic design professionals producing content for both print and digital media. What makes the tool unique is the enormous amount of time it saves by correcting hundreds of types of typographic errors. This means proofreaders no longer have to deal with dozens of awkward line breaks or missing spaces after initials. Adoption among editors and proofreaders will probably be slower, but we believe that once they see how easy the tool is to use, and that they can review and apply corrections individually, they'll come on board.
R
We also see promising applications in the digitization of historical books and documents, particularly in libraries and archives. When scanned materials are converted into digital text using optical character recognition (OCR), errors often occur that can be difficult to distinguish from imperfections in the original printing or historical typographic conventions. This is especially true of texts in foreign languages, where it may not be immediately clear whether a character is a dash, a hyphen, a quotation mark, or a misprinted punctuation mark. By drawing on its knowledge of language-specific typographic conventions, Typofix can flag potentially problematic passages and suggest corrections, making the otherwise time-consuming process of manual review significantly easier and faster for editors.
F
Another advantage is that the tool makes it easier to typeset languages the designer doesn't speak. Typesetters often apply the conventions of their own language to unfamiliar ones without realizing it. In French, for example, a space is required before the punctuation marks :, ;, ! and ?, while Portuguese commonly uses em dashes. Typofix naturally handles these differences. Its quotation mark correction features are another major time-saver. Anyone who has typeset texts in both American and British English knows that the two varieties often use primary and secondary quotation marks in reverse order. Typofix can switch between these conventions with a single click.




Workshop with international experts at the Klementinum.
Photo: Apolena Typltová
The project is based on expert knowledge and explicitly defined rules rather than artificial intelligence. Why did you choose this approach? What are its advantages and limitations, and what role do you think AI could play in future development?
R
When we were preparing the project, ChatGPT couldn't handle typography at all, and even during the project's development, the results were still extremely poor. From the outset, we knew we were creating a practical tool capable of detecting many errors, though not all of them. We therefore decided to use regular expressions, which work reliably and quickly, and run locally without sending any data to the cloud.
F
One limitation of this approach is that it doesn't work when the correct typographic treatment depends on context. For example, we still can't fully address the issue of percentages. In Czech, whether you write 25 % or 25% it depends on the context, which regular expressions cannot account for. AI can now handle this distinction, but exploring that possibility would require a separate project.
What do you consider to be the main scientific contribution of Typofix?
R
We succeeded in identifying and systematically categorizing different types of typographic errors, sometimes developing highly sophisticated solutions to correct them. We demonstrated that automated typographic correction is feasible, while also identifying the limitations of this approach. I also believe we've laid the groundwork for further research into typographic conventions, their systematization, and their conversion into "skills" that would enable large language models to apply these rules consistently and accurately.
Typofix could serve professionals in graphic design, typography, editing, and translation, as well as everyday users. How do you plan to evaluate its accuracy and practical usability? Have you also held discussions with companies such as Google or Adobe? And who will be responsible for maintaining and updating the database in the future?
F
For us, it's essential that the project continues beyond its initial development. That's why we designed it from the beginning with a degree of commercial viability in mind. We want to keep license prices affordable while generating enough revenue to sustain the project. We hope users will embrace the tool and that it will become a regular part of their workflow. Adobe, Google, and Microsoft are all aware of our project and have responded positively to it, but they're cautious about integrating it into their products. They want to see evidence that there's demand for the tool and that it works effectively. So far, our many discussions have provided valuable recommendations and interesting ideas, and perhaps one day we'll see Typofix officially integrated into a popular software application.






A preview of the publication Typofix Manual: Practical Rules for European Typography.
Graphic design: Konrád Karlík & Radek Sidun
The project is now entering its final phase, with the launch of a new application, the publication of a printed manual, and an international conference at UMPRUM. At the same time, you are preparing a follow-up European project. What have you achieved so far, and where would you like to take Typofix next?
F
We've achieved what we set out to do. The book has been printed, a new professional website is being developed, and the software tools will be available by the end of November. As we worked on the project, we realized it would be a shame not to continue. By focusing on "just" 24 languages, we've left out Turkish, Norwegian, Catalan, and several others, particularly smaller languages. And by incorporating AI, albeit with great caution and particular attention to sensitive data, we could potentially correct even more errors.
R
That's why we've applied for a Horizon Europe grant and assembled an excellent consortium. However, there are nearly 60 applications, and only two will receive funding. We'll be discussing all of this with our team during the November conference. Everyone is warmly invited! We'll share the story of how Typofix came together, and we'd love to welcome students, industry professionals, and anyone interested in the subject.
The conference, application launch, and manual launch will take place on Thursday, November 12, 2026, from 4:00 to 10:00 p.m. at the UMPRUM Technology Center on Mikulandská Street.
The event will feature a number of guest speakers:
Ben Wittner will present Multi-Scriptual, a project exploring the possibilities of cross-cultural graphic design and typography across different languages and writing systems. Henrik Birkvig will examine the origins, evolution, and design of the characters Ø/ø and Å/å as distinctive elements of Danish linguistic and cultural identity. Fraser Muggeridge will showcase experimental approaches to dynamic typography that challenge conventional typesetting rules. The program will conclude with Filip Blažek and Radek Sidun, who will introduce Typofix, a new publicly accessible database of typographic rules and a web application for automatically correcting typography in texts across all 24 official EU languages.
You can try Typofix at typofix.org
Interview conducted for UMPRUM Online by Michaela Režová.
F
Filip Blažek is a graphic designer, typographer, educator, and writer. Since the mid-1990s, he has focused on Czech type design and typography, lecturing on these subjects at UMPRUM as well as at numerous conferences in the Czech Republic and abroad. He is also active in graphic design through Designiq, a studio he founded in 2002. He is the author of Typokniha (2020, 2022) and Typomřížky (2022) and has contributed to numerous other books and periodicals.
R
Radek Sidun specializes in type design and typography, with a particular focus on diacritics used in languages around the world. At UMPRUM, he has served as co-head of the Studio of Type Design and Typography, head of the Graphic Design Department, and Vice-Rector for External Relations and International Cooperation. In 2012, he co-founded Briefcase Type Foundry with Tomáš Brousil. He is the author and co-author of several publications on typography and regularly lectures at design and typography festivals in the Czech Republic and abroad. He also works as an international consultant to type designers and type foundries.

This project is co-funded with state support from
the Technology Agency of the Czech Republic under the SIGMA Program.



‚tÿpøf¡x": More Than Just a Typographic Proofreader
Typographic conventions governing quotation marks, numerals, punctuation, and spacing reflect the cultural heritage of individual languages. However, globalization and standardized software often lead to these conventions being overlooked. In an interview with Filip Blažek and Radek Sidun, we explore the origins of Typofix, a project that combines international research with technological development. The project has produced a database of typographic rules for all 24 official languages of the EU, a printed handbook, and an application designed to preserve the unique typographic conventions of each language. The project was jointly supported by UMPRUM and the Technology Agency of the Czech Republic through its SIGMA program.

Filip Blažek and Radek Sidun.
Photo: Apolena Typltová
F
Filip Blažek
R
Radek Sidun
What was the original inspiration behind the Typofix project? Which of you came up with the idea first, and why?
R
We sometimes joke that it was laziness. After all, who wants to go through a text correcting dozens of typographic errors one by one, while also taking into account the language it's written in?
F
In fact, the idea grew out of our experience with a similar tool that we had developed and used successfully for almost 20 years when typesetting Czech and English texts. We felt it would be a shame not to develop a more comprehensive version that could support a wider range of languages.
One of the premises of the project is that globalization and the dominance of English in international communication are weakening language-specific typographic conventions. What do you think is lost when texts in different languages begin to follow the same rules?
F
Whether we like it or not, the English-speaking world has a significant influence on us. We encounter it everywhere. Just try typing Czech quotation marks in Google Docs, for example. Ignoring local conventions breeds mistrust and indifference, along with a largely justified sense that many global companies display ignorance and an attitude of superiority towards the rest of the world.
Can linguistic and typographic differences also be considered part of our cultural heritage? Why is it important to preserve them?
F
Absolutely! Typographic conventions carry traces of history, as well as cultural and commercial ties. They are passed down from generation to generation, and the professional community has worked to preserve them even through difficult times.
‚tÿpøf¡x":
„Dónde están los baños , por favor?“
«¿Dónde están los baños, por favor?»
Typofix
Examples
Before/After
Spanish
She said—I hope—“my ‘birdie,’ I love you.”
She said – I hope – ‘my “birdie”, I love you’.
Typofix
Examples
Before/After
English
La caissière du cinéma m ’ a recommandé un “film sensationnel”!
La caissière du cinéma m’a recommandé un « film sensationnel » !
Typofix
Examples
Before/After
French
The aim of the project was to document the typographic conventions of all 24 official languages of the European Union. How did you approach the research, and who did you collaborate with? Were there already comprehensive resources you could draw on, or did you have to establish and standardize the rules for some languages for the first time?
R
The lack of accessible information was one of the key factors that led to the creation of Typofix. How am I supposed to typeset a Bulgarian or Greek text correctly when the typographic rules for these languages are either unavailable or only documented in the languages themselves? And even when they are available in English, they are often full of errors or incorrect symbols, or different sources contradict one another. AI isn't much help either: one model claims there should be a space between a number and the % symbol, while another insists there shouldn't be. We realized that the only way forward was to work with experienced local typographers, typesetters, and educators who could help us establish a consistent set of rules.
A number of international experts contributed to the project. Could you describe the research process, how these experts were involved in collecting the data, and how the findings were put into practice?
F
Initially, we thought the experts would populate our database of rules themselves. But the deeper we got into the complexities of regular expressions, the clearer it became that most of them would find this too challenging. So we simplified the data collection process, creating tables that described a wide range of typographic conventions and asking the experts to select or provide the correct variants for their respective languages. We then generated test files over several rounds, which the experts reviewed and commented on.

Workshop with international experts at UMPRUM.
Photo: Apolena Typltová
Did you encounter any surprising differences or similarities between languages during your research? Was there a particular language that surprised you? Could you give a specific example that illustrates why typographic rules cannot simply be transferred from one language to another?
F
We discovered many unexpected conventions. For example, some languages don't have the concept of nested quotation marks and simply repeat the same symbols for a quotation within a quotation. In others, a dash is used in much the same way as parentheses, meaning that a closing dash is required even if the phrase appears at the end of a sentence. We also found that in several languages, years are treated as ordinal numbers and therefore always followed by a period, although using a period to indicate ordinal numbers is relatively uncommon among European languages. Another surprising discovery was that no two languages we analyzed followed exactly the same typographic rules. There were always differences of some kind.
Language standards evolve, and some conventions allow for multiple approaches depending on the context or type of publication. How did you account for this variation in a database intended to serve as a reliable foundation for automated corrections?
R
We addressed the issue of alternative conventions from the very beginning, since we know from Czech that opinions differ on things like the appropriate length of a dash. In some languages, users can only choose between different styles of quotation marks, while in others, they can select different spacing widths, thousands separators in numbers, preferred date formats, or whether short prepositions and conjunctions should be joined to the following word.
F
The entire system is designed to be highly flexible. For example, if someone requests the option to use a different preferred pair of nested quotation marks in Danish, we can implement it in just two minutes.
One of the project's main achievements is translating specialist knowledge into a database that software can process. How did you convert linguistic and typographic conventions into machine-readable rules? What was the most challenging aspect of this collaboration between typographers, language experts, and developers?
R
The conversion process itself was quite demanding, as we had to make sure that correcting existing errors wouldn't introduce new ones into the text. Rather than creating separate rules for each language, we developed a system of variables. These might include month names, abbreviations of books of the Bible, or conventions for writing musical keys, which differ considerably from Czech conventions in many countries. Another type of variable covers incorrect variants, such as the various forms of apostrophes, multiplication signs, or degree symbols. Once we had mastered writing even highly complex expressions, ChatGPT proved helpful, having finally learned how to create and modify regular expressions in 2026. In the end, however, the hardest part of the entire process wasn't converting the collected data into rules, but getting the experts to respond to our emails…


Typofix has the potential to serve not only typographers and graphic designers, but also translators, proofreaders, editors, and everyday users. Who is the tool primarily intended for, and how did the needs of these different groups influence its development? What features will it offer at launch, and what makes it unique?
F
Initially, Typofix will probably be used mainly by graphic design professionals producing content for both print and digital media. What makes the tool unique is the enormous amount of time it saves by correcting hundreds of types of typographic errors. This means proofreaders no longer have to deal with dozens of awkward line breaks or missing spaces after initials. Adoption among editors and proofreaders will probably be slower, but we believe that once they see how easy the tool is to use, and that they can review and apply corrections individually, they'll come on board.
R
We also see promising applications in the digitization of historical books and documents, particularly in libraries and archives. When scanned materials are converted into digital text using optical character recognition (OCR), errors often occur that can be difficult to distinguish from imperfections in the original printing or historical typographic conventions. This is especially true of texts in foreign languages, where it may not be immediately clear whether a character is a dash, a hyphen, a quotation mark, or a misprinted punctuation mark. By drawing on its knowledge of language-specific typographic conventions, Typofix can flag potentially problematic passages and suggest corrections, making the otherwise time-consuming process of manual review significantly easier and faster for editors.
F
Another advantage is that the tool makes it easier to typeset languages the designer doesn't speak. Typesetters often apply the conventions of their own language to unfamiliar ones without realizing it. In French, for example, a space is required before the punctuation marks :, ;, ! and ?, while Portuguese commonly uses em dashes. Typofix naturally handles these differences. Its quotation mark correction features are another major time-saver. Anyone who has typeset texts in both American and British English knows that the two varieties often use primary and secondary quotation marks in reverse order. Typofix can switch between these conventions with a single click.




Workshop with international experts at the Klementinum.
Photo: Apolena Typltová
The project is based on expert knowledge and explicitly defined rules rather than artificial intelligence. Why did you choose this approach? What are its advantages and limitations, and what role do you think AI could play in future development?
R
When we were preparing the project, ChatGPT couldn't handle typography at all, and even during the project's development, the results were still extremely poor. From the outset, we knew we were creating a practical tool capable of detecting many errors, though not all of them. We therefore decided to use regular expressions, which work reliably and quickly, and run locally without sending any data to the cloud.
F
One limitation of this approach is that it doesn't work when the correct typographic treatment depends on context. For example, we still can't fully address the issue of percentages. In Czech, whether you write 25 % or 25% it depends on the context, which regular expressions cannot account for. AI can now handle this distinction, but exploring that possibility would require a separate project.
What do you consider to be the main scientific contribution of Typofix?
R
We succeeded in identifying and systematically categorizing different types of typographic errors, sometimes developing highly sophisticated solutions to correct them. We demonstrated that automated typographic correction is feasible, while also identifying the limitations of this approach. I also believe we've laid the groundwork for further research into typographic conventions, their systematization, and their conversion into "skills" that would enable large language models to apply these rules consistently and accurately.
Typofix could serve professionals in graphic design, typography, editing, and translation, as well as everyday users. How do you plan to evaluate its accuracy and practical usability? Have you also held discussions with companies such as Google or Adobe? And who will be responsible for maintaining and updating the database in the future?
F
For us, it's essential that the project continues beyond its initial development. That's why we designed it from the beginning with a degree of commercial viability in mind. We want to keep license prices affordable while generating enough revenue to sustain the project. We hope users will embrace the tool and that it will become a regular part of their workflow. Adobe, Google, and Microsoft are all aware of our project and have responded positively to it, but they're cautious about integrating it into their products. They want to see evidence that there's demand for the tool and that it works effectively. So far, our many discussions have provided valuable recommendations and interesting ideas, and perhaps one day we'll see Typofix officially integrated into a popular software application.






A preview of the publication Typofix Manual: Practical Rules for European Typography.
Graphic design: Konrád Karlík & Radek Sidun
The project is now entering its final phase, with the launch of a new application, the publication of a printed manual, and an international conference at UMPRUM. At the same time, you are preparing a follow-up European project. What have you achieved so far, and where would you like to take Typofix next?
F
We've achieved what we set out to do. The book has been printed, a new professional website is being developed, and the software tools will be available by the end of November. As we worked on the project, we realized it would be a shame not to continue. By focusing on "just" 24 languages, we've left out Turkish, Norwegian, Catalan, and several others, particularly smaller languages. And by incorporating AI, albeit with great caution and particular attention to sensitive data, we could potentially correct even more errors.
R
That's why we've applied for a Horizon Europe grant and assembled an excellent consortium. However, there are nearly 60 applications, and only two will receive funding. We'll be discussing all of this with our team during the November conference. Everyone is warmly invited! We'll share the story of how Typofix came together, and we'd love to welcome students, industry professionals, and anyone interested in the subject.
The conference, application launch, and manual launch will take place on Thursday, November 12, 2026, from 4:00 to 10:00 p.m. at the UMPRUM Technology Center on Mikulandská Street.
The event will feature a number of guest speakers:
Ben Wittner will present Multi-Scriptual, a project exploring the possibilities of cross-cultural graphic design and typography across different languages and writing systems. Henrik Birkvig will examine the origins, evolution, and design of the characters Ø/ø and Å/å as distinctive elements of Danish linguistic and cultural identity. Fraser Muggeridge will showcase experimental approaches to dynamic typography that challenge conventional typesetting rules. The program will conclude with Filip Blažek and Radek Sidun, who will introduce Typofix, a new publicly accessible database of typographic rules and a web application for automatically correcting typography in texts across all 24 official EU languages.
You can try Typofix at typofix.org
Interview conducted for UMPRUM Online by Michaela Režová.
F
Filip Blažek is a graphic designer, typographer, educator, and writer. Since the mid-1990s, he has focused on Czech type design and typography, lecturing on these subjects at UMPRUM as well as at numerous conferences in the Czech Republic and abroad. He is also active in graphic design through Designiq, a studio he founded in 2002. He is the author of Typokniha (2020, 2022) and Typomřížky (2022) and has contributed to numerous other books and periodicals.
R
Radek Sidun specializes in type design and typography, with a particular focus on diacritics used in languages around the world. At UMPRUM, he has served as co-head of the Studio of Type Design and Typography, head of the Graphic Design Department, and Vice-Rector for External Relations and International Cooperation. In 2012, he co-founded Briefcase Type Foundry with Tomáš Brousil. He is the author and co-author of several publications on typography and regularly lectures at design and typography festivals in the Czech Republic and abroad. He also works as an international consultant to type designers and type foundries.

This project is co-funded with state support from
the Technology Agency of the Czech Republic under the SIGMA Program.