Diacritics.info,
the diacritical data

Yeelena De Bels (contact)
ANRT 2019 — 2021.

A digital platform gathering design information about diacritics from type designers giving advice on their native language(s).

In 2002, the Latin alphabet is used by 39% of the world population and constitutes 72% of the written production. However, this alphabet, consisting of 26 basic letters, fails to transcribe on its own all the languages which use it. Since the creation of the Latin alphabet, languages and their pronunciation evolved which prevents the Latin alphabet graphemes to write the whole oral languages and their new phonemes anymore. According to J.-C. Wells, a British phonetician, the contemporary languages use three solutions to keep using the Latin alphabet. 

The first one, found in written English, is the use of digraphs. Indeed, to overcome the lack of the strict use of the Latin alphabet, the English uses the combination of two graphemes to transcribe new phonemes. Thus, to translate the sound [ʃ] in writing, it is now possible to write sh by combining the letters s and h which independently match other phonemes. The second alternative, available in Icelandic, is the creation of new graphemes. These graphemes can be based on pre-existent letter shapes such as this Ð. The distinction between the new letters and some of the orthographic ligatures might be hard: is this glyph æ standing for a and e joined together or is it to be accepted as a letter on its own? The third option, which is the main topic of this project, is the addition of marks called diacritics on graphemes, graphemes being the smallest functional unit in a writing system (if you take the example of the Latin alphabet, letters are its graphemes).

These diacritical marks may appear above, below, through, or in the top right of a letter. They can also have different types of structure: they can be (or not) symmetrical, combined, they can use contextual variants (and change of shape regarding the letter they are added on, as the caron: č, š, ž / ť, ď, ľ ) or be connected to the letter. This adding of diacritics aims to counter the unequal quantity between the number of graphemes and the number of phonemes in a language using a writing system that fails to transcribe it on its own. Nowadays, the French include five diacritical marks. The acute modifies the phonetic value (é means [e]) when the value of the grave accent on the letter it is added on (è means [ɛ] whereas à has no phonetic value but allows to make the distinction between (la [the] et là [here]). The circumflex (â, ê, î, ô, û) has two features: opening the vowel on which it is added and/or indicating the elision of a letter (forest becomes forêt). The diaeresis can notify that two successive vowels are not being part of a digraph (mais/maïs) and the cedilla modifies the phonetic value of the c (ç means [s]). 

Nevertheless, while diacritics are existing for few centuries and have a real informative value, we can notice various incidents concerning them in the graphic and writing field. For instance, grave accents are often shown as a horizontal stroke, if they are even shown. Many capitals are missing diacritics when they should be accentuated. This irregularity, often considered as a facility, is criticized by a lot of professionals as graphic and type designers, linguists, editors, and writers. The typographic and linguistic world is therefore used to reactions of people who want to highlight the real diacritic needs as The Insects Project and How not to draw accents.

The diacritics.info project, described in this research report, is part of a global questioning about linguistic specificities and their influence on the functioning of different writing systems. Born from the conclusions brought during the writing of a thesis dedicated to French diacritics, this project aims to answer a flagrant lack of typographic resources dedicated to diacritics. It gradually took the form of an informative platform focused on the relationship between diacritical marks, languages, and typographic design. Its first layouts, proposed in 2018, showed an online tool listing information on the history, design, and use of diacritical marks.

When, how, and why did diacritical marks appear in some languages while they were absent in the English language? How did their use evolve in the course of our history? Why did the way they were drawn differ from one language to another? At that time, still referred to as latindiacritics.eu, this project was intended to answer all these questions and much more. It is now obvious that it was both too ambitious and too indecisive. By not putting any content aside, potential users of the tool would be overwhelmed with information and the function of the platform would be rendered unreadable.

After an enforced pause, the project was restarted a year later, when the opportunity to be studied and built at ANRT presented itself. Indeed, the ANRT seemed to be the ideal place to put into perspective the choices previously made. A tutored follow-up would allow the project to come to an end without being overwhelmed by the number of elements to be thought of, gathered, sorted, and arranged. To be a functional tool, diacritics.info needed the time, the space, and the supervision offered by the Atelier National de Recherche Typographique. Far from starting from scratch, the project was to benefit from a complete overhaul with the help of the research and reflection undertaken during the DSAA.

As presented in 2018, latindiacritics.eu was defined as an informative platform dedicated to the diacritics of European languages using the Latin alphabet. This platform, still in the project stage, was then divided into three parts (About a specific diacritic, About a specific language, Find a typeface), each dedicated to providing different information on the subject of diacritics.

• The first part, referred to above as About a specific diacritic, provided textual and visual information about the design of each diacritic sign listed.
• The second part, entitled About a specific language, allowed the user to access the history of each language listed, through the prism of diacritics. Why, when, and how did they appear in the writing of the selected language? Why, when, and how did they evolve afterward?
• The third and last part, Find a typeface, was a tool to find, among a proposed list, the character or characters compatible with the selected language or languages.

Each of these parts was visually subdivided into three sections in the form of columns. The first column, on the left, was intended as a selection menu. Thus, it allowed access to the content of a diacritic or of one or more languages. The central column was dedicated to the textual contents: from the drawing advice for the first part to the list of characters compatible with one or several languages for the last part, passing by the history of the diacritics of each language in the second part. The third and last column, on the far right, contains iconographic content to complement the textual content.

In order to ensure the quality and credibility of the information published on the platform, the content related to languages and the design of diacritical marks had to be written by experienced type designers, contacted to speak about their native language. This method made it possible to avoid relaying urban legends while collecting drawing indications from real speakers and readers of each language.

While initially it had been envisaged to dedicate the 18 months of the post-master to the aggregation and re-design of the platform, the need to review the basics of the project was quickly felt. Indeed, it would have been counterproductive to launch into the writing and design of a content whose very definition had to be reviewed. The time covering the period from October 2019 to February 2020 was therefore dedicated to an analysis re-questioning each of the choices previously made. This questioning phase unfortunately proved to be not very effective. Indeed, by constantly bringing new questions, new research, this stage contributed more to widen the field of possibilities than to circumscribe the project. The fact is that the project lacked a clearly defined purpose and function. The absence of a real guideline did not allow for an effective distinction between useful and potentially superfluous questions.

Nevertheless, among the multitude of reflections undertaken, two proved to be fruitful for the continuation of my research: the one on the selection criteria of the languages and writing systems to be treated and the one on the way to aggregate and create the content of the platform.

This project is intended to bring information on the linguistic specificities present in our writing systems, it was not very coherent to limit itself definitively to the European languages using the Latin alphabet. The drafting of a list of all languages with diacritics was undertaken in order to determine the criteria on which to base the selection of languages to be addressed. Complications were quickly encountered: how to distinguish between forms specific to one or more graphemes (such as orthographic ligatures) and diacritical marks? How to distinguish languages using accents from languages using a second transcription system? An eye not used to reading kanji and kanas might be tempted to interpret furigana as diacritical marks when they are closer to phonetic transcription. Sorting out the more than 3,000 languages spoken and the more than 150 writing systems would require an incredibly time-consuming amount of research. The choice was therefore made to limit the first version to languages using the Latin alphabet and to focus particularly on the official living languages while keeping the door open to other writing systems for future versions.

Once these initial criteria were established, the question of content creation arose. As mentioned earlier, the authorship of the content of a given language was envisioned as being the responsibility of a “confirmed” typographer who speaks and reads that language. This method would have required the establishment of a rather strict protocol in terms of making contact, marking out the different themes to be addressed, the precision and depth of the information provided as well as a significant amount of time for proofreading and corrections. Given the complexity, as a professional, of responding to all the calls for projects, it would have been somewhat absurd to ask these same professionals to dedicate an inevitably long time to the research and writing of such specific content, which is moreover “free”. It was also difficult to restrict the choice of one author per language when we know that reading and drawing habits vary from one person to another. The solution quickly came in the form of modular questionnaires to be sent to a certain number of designers of each language, thus allowing to collect a large number of answers to obtain enough material to study and synthesize.

As described earlier, this project was initiated as a result of various problems identified during the writing of a thesis dedicated to French diacritics. In particular, three problems were identified:

• lack of knowledge of the languages, their individual histories, and their internal norms of use and writing;
• shortcomings of a certain number of characters when it comes to diacritical signs (incorrect design, absence or interchanging of signs and encoding errors);
• difficulties in transcribing and transmitting certain diacritical signs (limitations due to physical and digital tools).

Driven by the desire to respond to these problems, this project encountered some difficulties in positioning itself and determining precise objectives, and therefore in making choices. This dense research has generated abundant and stimulating questioning but it was necessary, to structure things, to release a time of exercise intended to better define the ambitions of the project.

The first study was undertaken in order to better define the objectives of the project consisted of compiling information concerning a particular diacritical sign, the superscript dot. The research concerning this sign focused on two topics in particular:

• the first appearances and the evolution of form and usage;
• the various typographical problems still encountered.

The first traces of this sign, belonging to the category of dot accents, can be observed in some medieval Latin manuscripts of the thirteenth century. Gothic writing and its particular rhythm produce letters with very similar appearances, greatly slowing down the deciphering of certain pairs of letters (im / un) readable in a juxtaposed way. The dot on the i, having at the time more the appearance of an acute accent or a crescent moon, helped to distinguish the letter i. This sign, sometimes called tittle, in particular, because its diacritical function is not always active. In French, for example, it is used in a non-diacritical way and is more to be considered as an integral part of the lowercase variant of the letters i and j. Its use does not bring any phonetic or linguistic information. In Turkish, on the other hand, the dot on the i has real linguistic importance and has a diacritical function, especially in the use of ı alone. The fact that the same visual sign, here the dot over the i, can be and not be a diacritic is quite revealing of the complexity of defining and understanding diacritics. 

The design of the superscript dot may encounter three difficulties and peculiarities:

• case-by-case alignment depending on the letter on which it is placed, especially on letters such as the d or the r;
• a variable shape, especially in typographic characters simulating handwriting and those with strong calligraphic references;
• the need to create conditional ligatures to avoid an unfortunate collision with a neighboring ascender.

A last curiosity concerning the overprinted dot can sometimes be encountered when using certain characters: the tittle and the umlaut, although both composed of dots, do not always share the same shape. It happens that the dot on the i is square and the umlaut is composed of round dots and vice versa. 

The points of interest defined by this study have been evaluated from information collected among the writings of authors who feel concerned by diacritical marks and who have already undertaken quite profound research on this subject. From this observation arises a new question: how to choose a subject without knowing precisely what the need is? What are the questions asked by “laymen”?

Questioning the Questions

In order to determine the most common questions regarding diacritics, a second study dedicated to the collection and analysis of questions was then undertaken. This collection took place on the Quora platform, previously used to collect part of the information gathered during the first study. This platform allows everyone to ask and answer questions while highlighting the type of profile from which the information comes.

A little less than a hundred questions on the subject of diacritical marks have been gathered, duplicates and subjective questions having been discarded. An initial analysis of all the questions collected revealed 17 types of subjects, thus making it possible to understand which themes were most frequently addressed. First, the limitations of digital media and the rules for the use of certain diacritical marks were questioned, followed by questions related to the history and importance of diacritical marks.

In combination with the first analysis, a second examination dividing the questions into different categories and subcategories of topics allowed us to define four main research themes:

• definition and nuances of the term “diacritical sign”;
• information about languages and their diacritical marks;
• information about diacritical marks signs (meaning, use, and form);
• information about the limitations of digital media (keyboards, software, OS, and computer languages).

The two studies previously conducted, while not revealing an obvious purpose for this project, identified three possibilities for editorial content:

• content focused on the popularization of the definition and nuances of the sign, difference with other terms, statistical figures, and uses;
• content focused on the history of languages through the prism of diacritical marks: contextual information and character sets, the context of the appearance of diacritics and diacritical marks and the evolution of their form and use;
• a content centered on the diversity of design of diacritical marks in our contemporary languages: contextual information and encoding, drawing tips.

These three options, which can already be found in another form in the previous research of this project, each require their own research time as their subject is so vast. Without real surprise, these studies have finally accentuated the need to make a choice and to define a precise and consistent guideline for this project. In order to determine the objective of this project, wouldn’t it make sense to go back to its beginnings? What were the conclusions of this thesis that gave rise to all this research?

Don’t some of the problems we still encounter today in the use of our diacritical marks come from the lack of complete and didactic resources on the design of diacritical marks? How can graphic designers and amateurs have high standards in terms of aesthetics and typographic syntax, whatever the language they are designing, if their tools are incomplete and incorrect? How can we ask the creators of our contemporary typefaces to produce complete and efficient tools for diacritical marks if the resources on this subject are almost non-existent and often incomplete?

In short, wouldn’t the goal of this project be to make available information on diacritical mark design in order to feed the practice of type designers and thus allow them to complete their knowledge on the subject? Based on the premise that designers are responsible for the tools they create and intend to use, this project is intended to take the form of a digital tool that gathers and organizes the information necessary for the design and development of more complete and better-constructed diacritics.

To define the type of content to be gathered on diacritics.info, it was necessary to understand more precisely which types of profiles would consult the platform and what their needs would be. Among the designers, three types of profiles could be defined:

• type designers wishing to make their typeface compatible, in terms of the character set, with a given language;
• type designers wishing to obtain design advice on a diacritic or diacritics;
• type designers wishing to make their typeface compatible with one or more languages, both in terms of type designers wishing to make their typeface compatible with one or more languages, both in terms of typeface set and design.

Once these three profiles were established, it was then possible to list the essential information to collect and present on the platform:

• each diacritic had to be associated with a list of combined glyphs, and for each of these combined glyphs the languages using them in their writing system had to be indicated;
• each diacritic and combined glyph had to be associated with one or more Unicode codes (Unicode code of the diacritic and the Unicode code of the combined glyphs in lowercase and capital letters) to facilitate their encoding;
• each diacritic had to be accompanied by recommendations in terms of design with details on the linguistic variants.

With this list finally clarified, the question of content aggregation arose.

Drafting & Illustration of Questionnaires

To provide access to the most qualitative information possible, the need to interview designers of each language arose again, and with it the need to create questionnaires for each language. A test version of the questionnaires was first made using the spreadsheet tool Google Sheet. The idea was to enhance the collaborative spirit by allowing the document to be built, completed, and refined as contributions were made. This participative document format was intended to make the designers feel comfortable by showing them that they were not alone in contributing their point of view and information. However, it quickly became obvious that this tool could have the opposite effect, as the typographer could feel intimidated by this document readable by all.

The decision was made to change the tool for a simpler and less intimidating one: Google Form. With this new format came the possibility to use illustrations, which had not been considered with the previous tool. This new opportunity has allowed us to clarify certain questions and thus to gain precision in the answers. For example, in the case of the slope of the acute accent, it is easier to choose between various illustrated options in which our eye immediately spots the incongruous/unusual forms, and to complete or qualify this choice with a short comment, than to decide between options only expressed in degrees of slope.

When the questionnaire was first sent out, the illustrations were composed of characters that were sometimes modified to illustrate each of the options. For example, the slope of the acute accent in the serif and sans typefaces had been altered to propose a 20° slope, a 30° slope, a 40° slope, and a 50° slope, without fully reworking the balance of the accent. Not surprisingly, the questionnaire responses quickly showed that this approach was problematic: it was more difficult for the designers to determine which slope was most familiar or most pleasing to read when three of the four options were inconsistent. To overcome this problem, the decision was made to look for concrete examples in serif and/or sans type for each situation.

This new technique was quite time-consuming, as each questionnaire required reviewing at least part of the illustrations. Indeed, even languages using common diacritics do not necessarily use the same combined glyphs: illustrating the options of questions on the acute accent could not be done in the same way for French and Croatian, as French uses the acute accent on vowels and Croatian on a consonant. Similarly, some questions were not appropriate in certain linguistic contexts: although the umlaut is used in Estonian, the ï is not part of its writing system, so all the questions relating to this specific case of usage were useless.

The realization of these questionnaires and the collection and synthesis of the answers received allowed us to define two categories of elements:

• the forms intrinsic to a given diacritical sign in a more or less specific linguistic context;
• the intrinsic forms of a given typographic character, determined by its historical references, its use, etc.

The differences between these two categories of elements having been defined from a questionnaire focused on text characters, it is important to note that their limits are quite different in the field of title characters. Thus, the cedilla stick, which is often observed in display typefaces, is not disturbing when it is used on poster titles or a city name like Besançon for example. However, it can be much more problematic when used to compose a literary work: the body of the text is not the same, the reading time is also different and what can be pleasant in a punctual use can quickly be disturbing and tiring for the good understanding of a text.

The reading of the questionnaire has sometimes disconcerted the designers who answered it, as it may seem absurd to define the shape of a sign as something fixed. It is precisely the diversity of the answers to this questionnaire that allowed and still allows us to determine what is the margin of maneuver between the respect of the reading habits linked to a language, a culture, and the stylistic and aesthetic liberties that we wish to take.

Digital Resources

In addition to the answers to the questionnaires, three websites proved to be very informative. For example, the Charset Builder tool created by Alphabet Type was a great help to build the first part of each questionnaire, dedicated to the verification of the diacritical signs used in each language. This tool lists every sign used in all the languages listed in many writing systems. It allows designers to create a list of signs according to their criteria and then export it in different forms: glyph representation (ÿ), Unicode hexadecimal code (00FF), glyph name (ydieresis). This choice of a unique resource to list the combined diacritics and glyphs by language was influenced by the fact that the information was too divergent according to the different resources available on the subject.

With the same concern for optimizing search time, the information related to Vietnamese and Polish diacritics was gathered entirely (or almost entirely) from two digital resources (Vietnamese Typography by Donny Trương & Polish Diacritics, How to? by Adam Twardoch) that proved to be complete enough to be considered as primary sources. Before using the content of these two platforms, it was necessary to check the quality of the information gathered there through two graphic and type designers: Thy Hà, herself the originator of a research project on Vietnamese diacritical marks, and Michał Jarociński co-creator of Typoteka and typographer at Dada Studio. Both resources having been validated and completed by digital correspondence, the information about Vietnamese and Polish diacritical marks was thus mainly collected from these two websites.

The old version of diacritics.info, then under the title latindiacritics.eu, gave access in the first place to textual information, both to inform about the history of the diacritical marks and to give advice on their design. The diagrams and illustrations came only in a second time and had then more of a supporting role. The focus of the project, since refocused on the profile of the typographer, has highlighted the need to concentrate most of the information gathered previously within schematic illustrations. These illustrations will then be complemented by succinct textual information (using interactive bullets) sometimes necessary to clarify the illustrated subject.

Between May 12th and July 7th, James Edmonson published a series of illustrations on behalf of his company, OH no Type Co. The purpose of this series was to popularize certain optical principles necessary for the drawing of Latin letters. These publications served as the main reference for the creation of the first drafts of illustrations for the diacritics.info platform. These illustrations are not very busy, they have a rather relaxed atmosphere and gently draw the eye to certain rudiments of typographic design without locking the reader into a set of absolute rules.

Depending on the typographer’s needs, two main types of information can be searched for: the typeface set specific to one or several languages and design recommendations for one or several given diacritics. To do his research, the typographer could have in his possession one or more information from the following list:

• name of the language (with which he wishes to make his character compatible and/or in which the desired diacritic is used);
• grapheme/basic character on which the diacritic is used;
• types of shape present in the design of the diacritic (round, angular, straight, irregular/sinuous);
• types of structure (symmetrical, one or more diacritics, use of variants, connection with the letter);
• types of vertical alignment (above, below, across, top right).

In order to allow the type designers to access the result of their search, this information can be used as a filter and sorting parameters. The list of diacritics compatible with the search will be filtered in this order:

• will be displayed the diacritics used in the selected language(s);
• will be shown the diacritics that can be added to one of the selected graphemes;
• will be present the diacritics that have at least one of the activated form features;
• the diacritics structured according to the activated options;
• will be visible the diacritics of which at least one of the variants corresponds to the selected alignment.

Following this sorting, the combined glyphs matching the search will be highlighted to allow type designers to identify which combined glyphs are compatible with the searched character set. Via each diacritic, type designers will then be able to access a set of visual recommendations, sorted by categories: shape, height, width, fat, connection, endings, alignment, spacing.

They will be able to fly over all the themes or access a specific theme using a chaptering.

A long time was dedicated to the analysis of a large number of digital platforms coming from various universes. This study has gradually refocused on three categories of sites: sites selling and/or displaying typefaces; digital tools (online or local) dedicated to the creation and manipulation of typography; and sites dedicated to the publication of research texts and/or articles more or less focused on (typo)graphic design. As a good understanding of a tool depends partly on the coherence of the codes used within an ecosystem, it was important to fully anchor the platform in the visual universe of the typographic world.

After a few quick sketches, the need to use prototyping software was felt. At first, learning Adobe XD software seemed logical enough, as it was part of the range of software offered by Adobe. At the time, it seemed easier to use, its overall functioning being very close to the trio already in daily use (Photoshop, Illustrator, and InDesign).

The first drafts of the functional model were made on Adobe XD. The filter menu, envisaged from the start, was then much more complex and was divided into three parts, themselves composed of several parameters:

• Linguistic characteristics (scripts, languages, graphemes);
• Structural characteristics (shape, symmetry, number of diacritics, use of variants);
• Positioning characteristics (vertical alignment, connection).

As a result of discussions related to the handling of the filters and after several tests on different potential users, the specifications were modified. Three decisions stand out from these modifications:

• moving from Adobe XD to Figma a web app allowing a more fluid and efficient working time with a more precise result;
• simplification of the menu structure by rewording some names and removing categories;
• creation of an additional search tool, to introduce the different filters and to help with their handling.

This tool, now called the Search tool, allows the user to build, step by step, a question formulating his search. It has been partly inspired by the Indian Type Foundry website search function.

Undertaking and completing a research project is far from easy, and in any case, it is much more complex than one might imagine. It is sometimes complicated to find one’s way between the various desires and ambitions one arrives with, the curiosity one learns to develop towards the subject one is exploring, and the various needs one wishes to meet. Choosing to close the door on certain possibilities can be frustrating and disappointing and it can sometimes feel like giving up on certain needs. There is something special about the evolution of a research project, and the final answers will likely take on quite different forms when carried out by different people at different times. This year’s research time was heavily impacted by a global pandemic. The scarcity of social relationships, even more, intense when a research project is undertaken alone, the lack of a clearly defined workplace, and the dissolution of the original energy into a daily routine nearly killed this project. Thanks to the unfailing support of our tutors, motivation and desire fortunately resurfaced.

During this time of research, it was necessary to prioritize certain explorations and the choice was made to focus essentially on the reading and drawing habits of the diacritical marks within each language. Inevitably, two other potential research topics were discarded: investigations concerning the different functions and uses of diacritics, and the history and evolution of diacritics. These two topics, still essential to our full understanding of the subject, could be the subject of future research.

Still in the mock-up stage after four years of research, diacritics.info will soon see the light of day in the form of a web platform intended for type designers wishing to improve the compatibility of their typeface, both in terms of typeface set and design, with one or several languages. With design recommendations, this tool will allow any professional of the typographic world to support his knowledge in the matter. Although the first version of this site is not yet published, the reflections are already going well to determine what will be the next step of the project.

Shortly, two areas will be further developed: the improvement of the tools offered to help with the typographic creation of diacritical marks and the addition of functionalities related to the completion and augmentation of content.

Initially, priority will probably be given to adding new features to improve the typographic tools already available. For example, it is planned to increase the feature allowing to build one’s own character set: following the example of the tool created by Alphabet Type, diacritics.info will allow, in a later version, users to export/copy and paste the character set obtained thanks to the filters. In the same continuity, it is now envisaged to make it possible to save certain contents (list of signs, drawing recommendations, etc.), and thus to gather them in groups and sub-groups.

Later on, a modular questionnaire will be added, based on the structure of those currently used, as well as the possibility to propose modifications or corrections of the design recommendations. It might also be desirable to add a more human dimension based on the sharing of historical or vernacular resources. These, in addition to federating a community around this subject, will allow to decomplex the type designers by showing them the extent of the possibilities in terms of the typographic design of diacritics. In short, this would add to the platform a more concrete, more open aspect with an iconography less abstract than schematic illustrations.