Languages, scripts, and typefaces (2006)

[Response published in tipoGrafica no 70 (2006)

Do you consider that new technologies will enable those languages that have historically not been represented in font design, to incorporate the sounds appropriate to their tongues?

Hm… the question is somewhat misleading. The ‘languages that have historically not been represented in font design’ bit suggests that typeface designers who are native users of the other languages, the ones that have ‘historically not been represented in font design”, designed their typefaces with the sounds of their language in mind. This is not the same as saying ‘I can hear the words when I read’ or something like that; it means that the designer would have specific sounds in mind when designing a particular glyph. I’m pretty certain this is not the case; even more, I think the hypothesis runs counter to the basic mechanics of natural scripts.

Are fonts developed for specific languages? Even in the old days of 8-bit codepages, when each font file could take up to 256 characters, any single font allowed many languages to be typeset; the same font would do for English, Spanish, French, Finnish, and Italian, just as the same font with the declaration ‘CE’ following its name would cover Croatian, Czech, Estonian, Hungarian, Latvian, Lithuanian, Polish, Romanian, Latin-based Serbian, Slovak, Slovenian and Turkish (I think that’s all).

Such groupings (and there were plenty) were a combination of technical limitations (fitting all the characters in the font file) and commercial expediency: development, distribution, and retail channels. Each of these fonts claimed it could be used to typeset all these languages – and it did, offering a more or less adequate typographic representation of a script’s subset. I am choosing my words carefully here, because the point I am making is that typefaces offer interpretations of scripts, not languages.

We can shed some light on this distinction if we consider another distinction. So far I’ve been using the term ‘character’, but in fact this is not strictly correct. At the heart of contemporary applications and typefaces is the Unicode standard: a system for assigning a unique identifier to each character in any script ever used by humans. In this sense, ‘Latin small letter a’ and ‘Greek small letter alpha’ are characters, but ‘a’ and ‘α’ are not: they are glyphs: typographic representations of characters. In other words, all ‘α’s in all the typefaces in the world are the same character: Greek alphas, and all the ‘a’s are Latin [script] ays (how do you spell ‘a’?) – not English or Spanish or Finnish ays. To put it bluntly: the character implies a specification for the formal configuration of the glyph (relationship of positive and negative spaces) but is ignorant of the specific shape.

The relationship between character and glyph is, in my view, strongly analogous to that of a glyph and its voicing within a language. The Latin ‘a’ implies an ‘envelope’ of sounds within each language that is written with the Latin script, and a set of relationships of this sound with neighbouring glyphs. The leeway in speaking the glyph is, however, considerable; even to my unfamiliar ears a word such as ‘tipografia’ sounds very different when spoken by my Argentinian, Mexican, or Spanish students. Should they be writing with different glyphs for the ‘a’ in each language?

If, for the sake of argument, we posited that: yes, each of these three languages requires a different ‘a’ (or a different combination of ‘gr’, for that matter) then we must automatically decide what is the minimum difference in enunciation between the two variants that will trigger a choice one way or the other. Do we document the range of possible sounds that pass for ‘a’ in speech in each of these languages? This can quite quickly turn into a massive exercise in mapping speech patterns and deviations – the age-old classification problem of the infinite pigeonholes, the ‘duck-billed platypus’.

I can almost hear you say: ‘hold on there, you’ve missed the point! We should only be looking at each language in turn, not compare across languages!’ OK; but what will we do with dialects, regional variations, inflections dependant on age, social class, education level, professional affiliations, and the like? Again, this is a dead-end. Should I write English with different glyphs from my children? I have an expat’s accent, forever getting the length of vowels and the strength of consonants wrong; but my children, who go to nursery with all the other children in the area, speak English with an impeccable accent (so much so, they already correct their parents…).

There is only one area where we can strive for a close, one-to-one relationship between spoken sounds and the glyphs of a typeface, and that is the domain of linguists who document spoken language. (The International Phonetic Alphabet is fairly comprehensive in its coverage of sounds the human larynx can produce, and only extended when someone researching vanishing or spoken-only languages come across a new sound.)

Going down that route will take us further away from the visible form of language, and into questions that deflect from and confuse the study of typeface design; this must, by definition, be limited to the representation of characters, not of sounds. The formal qualities of the glyphs may bear many influences, from the direct (mark-making tools such as nibs and brushes) to the lofty (theories of construction); and they will normally take a long time to define a envelope of expression for each time and place (the strength of which is tested each time we come across the ‘design-by-dictat’ approach seen in Korea, Russia, Turkey, and – most recently – in the ex-Soviet states of the Caspian).

So what about the bells and whistles? Current technology promises a range of options that were not available before outside specialised environments. These must be seen as limited to the level of enriching the typographic expression of a script, but not jumping out at the level of the sounds the glyphs will generate in specific users. So, if a Polish typographer prefers diacritics to lie flat over the vowels, whereas a French one may opt for the upright ones, all the better if the font files can provide both, change from one to the other on the fly, and keep both users happy. Similarly, if the Dutch have a mark that looks like an ‘i’ and a ‘j’ close together and are taught at school this is one letter, you would hope that the whole chain of text editors, word processors, spell checkers, dictionaries and thesauri would recognise it as such. Speaking is interpreting sounds within a culturally-sensitive envelope; so is typeface design: defining glyphs within the acceptable spectrum of each character. But the designer cannot interfere where there is not linguistic ground to step on: if it means different things, go ahead and make a new glyph – but if it just sounds different, well, that’s just a reflection of the imprecise, fluid, and constantly mutable nature of human expression.

Let’s stick to mark-making, I say.