by | Apr 29, 2024


Against the tyranny of tech progress, my latest fonts wring new design concepts from a 20-year old innovation: contextual alternates

In a three-act career as art director, graphic designer and subsequently type designer, I’ve been through many paradigm shifts, of varying severity. After missing an upgrade or two, and with a dose of planned obsolesence, it certainly does seem like a paradigm shift has occurred—when even the simplest things don’t look and work like they used to. And then every so often, and with increasing frequency, along comes a real sea change, like digitization, the internet, social media and AI.

The exceptionally rapid deployment of AI, when we have yet to adjust to and rein in the disaster of social media, has prompted my interest in an area of design made possible by a previous generation of design tool. Slowing down, not throwing out the existing tools when they still work perfectly well. No, not going quite as far back as pen and ink, although that’s still viable. First something else.

I began using the contextual alternates feature of OpenType in 2005, shortly after it was introduced, and explored its possibilities for a good ten years before losing interest. But now, with the passage of time opening up some perspective, and with the attraction of it being an alternative to the grindstone of new tech, I’m developing a fresh idea of what may be done with contextuality—such as in the headline above, in which roman and italic characters flip, in a single font, Scotch Biplex.

Here are the generations, not of technology, but of an avenue of design it opened up, code-name: <calt>.

The raison d’être of the Contextual Alternates feature was to enable script font typography to assume the complex construction of cursive writing, and this is how it was registered by Adobe in the OpenType spec: “In specified situations, replaces the default glyphs with alternate forms which provide better joining behavior…” And that is all, no mention of other uses such as those detailed below, which are unintended consequences. Transferring the traditional joining rules of sophisticated calligraphy into digital encoding—a sort of “dynamic in-app ligaturing”— Adobe’s Robert Slimbach introduced Caflisch Pro (2001), with the layout application InDesign supporting it. Then FontLab emerged as the font-making tool that enabled independents such as Christian Robertson (Dear Sarah) and myself to also employ the feature. I upgraded Handsome (above) to OpenType in 2006, adding contextuality.

Recognizing that there are many possible formal relationships between adjacent glyphs, beyond cursive joining style or ligatures, my next foray into contextuality addressed something different: proximity. I created a rounded sans serif font, Softmachine (2006, above), in which the closest distance between letter parts—both within and between letters—is equalized. Kerning alone cannot accomplish this neatly, without the many character alternates that OpenType permits; note the two t’s above.

LettError’s Beowolf (1989) and Kozmik (1993) introduced the notion of randomness to digital typography, challenging the convention that every time a character repeats, the glyph should be identical. However, the non-standard programming in those typefaces was not widely supported in layout applications. Then in 2005 several pseudo-random types were released with OpenType contextual programming by Tal Leming: Ken Barber’s Studio Sable and Studio Swing, and Christian Schwartz’s Local Gothic, which mixed letters of different horizontal scale. Since then many foundries have followed suit. Patrick Griffin of Canada Type weighed in with Chapter 11 (2009) and Outcast (2010), distressed styles in which disparate alternates suggest the randomness of wear and tear, and uneven, haphazard printing.

In 2008, I designed Duffy Script (above), an interpretation of the lettering of Toronto illustrator Amanda Duffy.  Each weight contains four glyphs for every character (including all numbers, punctuation and symbols). These are coded to set in a non-repetitive order, for a subtle, natural effect—and this is generally what is meant by “random” in OpenType fonts. The variant glyphs for each character are not radically different, but consistent in the way that letters from Duffy’s hand exhibit slight modulations from a distinctive pattern. Ironically, in a truly random font, adjacent repetition of identical glyphs would occasionally occur, but this is something that the pseudo-random designer steers clear of.

In 2010, for the OpenType makeover of Fontesque (originally designed in 1994), I again applied a pseudo-random feature, appropriate for a design with a pointedly hand-drawn aesthetic. And in 2016, I employed contextuality to create a complex bounced effect in Gambado, in which the alternate glyphs have the same outline as the default, but with a different baseline shift and rotation.

With Neology (2004), I addressed readability through type design as empirical process, by isolating letter form as a variable. That is to say, fundamental “topographic” letter construction, not proportion or curve shape. I wanted to see what effect such variation in letter form could have on readability, whether absolute conformity of character shape has any performance benefits. Neology mixes geometric and grotesque variants of the same basic style, via pseudo-random code.

If the essence of contextuality is providing two or more different appearances for the same character, just how different might they be? Up to now, in the interests of fluent readability, I had assumed that the variant glyphs should be closely related in style, but why not really push it? 

Perhaps I was prompted by the death of Jamie Reid in 2023, and recalled his “ransom note” typography for the Sex Pistols in the late 1970s. At any rate, it occurred to me that I could use the four-alternates coding I’d designed for Duffy Script to create that kind of effect, hence Polypraxia (above, 2024). It was, of course, not quite so simple as I had expected—in order for whatever combination of letters being set in the font to present even colour, the four styles have to strike a balance: sufficiently related to create an even flow, and yet with enough sustained variety that no sequence of glyphs appears to all belong to the same style, or has the odd glyph jump out as markedly different from the rest. Thus the four styles are the same bold weight, in (1) a condensed didone, (2) a condensed grotesque, (3) a humanist sans, and (4) a caps-with-small-caps geometric sans. It would seem that for typography some fluency is required, even in a “disfluent” font. As ever, a balance of unity and variety.

Subsequent to Duffy Script, I had stopped making four-variant fonts, settling on two-, the theory being that the narrow arc of saccadic fixations does not permit the reader to recognize structural dissimilarities in characters outside each visual “grab,” with one‘s attention focused on decoding the text, not playing spot-the-difference with the typography. In Scotch Biplex (top of post, 2024), I pursued a more rigorous application of extreme stylistic contrast, mixing the Roman and Italic of my Scotch Modern. This genre was, I intuited, the best bet for alternating roman and italic, with the very obvious disparity of construction (not least the 20° of slant difference) balanced against the even colour provided by both the similarity of widths between the component styles, and the enormous pot-hook serifs of the italic lower case, which fill out the mid-x area. Strangely, the angle of italic slant matches the roman diagonals, as can be seen in the title of this post. It was meant to be.