Andrea Stanton
Typewriters and computers were designed with English and other left-to-right-scripts in mind, while Arabic is written right to left. As a result, technicians have described adapting these technologies to Arabic script as a particularly thorny challenge. But behind the technical problems of typing in Arabic (or Persian or Urdu) stands a world on fire—and not the world of the Middle East. Our assumptions about technology and its relationship to the Roman alphabet, and to the unidirectional scripts it supports, have led us to ask the wrong questions—and to seek the wrong solutions. Why has typing in Arabic been considered such a technical challenge, and why are the solutions—past and present—so inadequate? Why do users who work in multiple languages and scripts find it so difficult to produce legible word-processed text in Arabic and English, or any other combination of Arabic-script and Roman-script languages? Why are the solutions so technical? Arabic is not the problem—thinking of Arabic script as a problem is.
At heart, typing in Arabic—especially in a document that also has Roman-script text—is not a technical issue and needs more than a technical solution. Thinking of Arabic-script typing as a technical problem is the result of societal assumptions made long before the computer and word processing were developed—assumptions that shaped the development of Arabic-script typewriters. Those assumptions are, first, that Arabic script was a particularly “challenging” one that required modifications in order to be used on a typewriter, and, second, that typists and word processors would work in one script at a time. Asking the wrong questions has led us to seek partial, technical solutions. Asking better questions can help us address the culturally laden assumptions about unidirectional script as a typing norm and embrace the reality of a global world in which many people live and work across multiple script forms.
Both the typewriter and the word-processing component of the computer first developed in the United States. They were conceived, designed, prototyped, manufactured, marketed, and distributed in English, and with English in mind. The standards that developed with them reflected that fact: Roman script, unidirectional left-to-right, no accents, and no cursive. Yet from the typewriter’s mass production in the 1870s, the companies that produced it, the consular officials who promoted American products overseas, and entrepreneurial inventors around the world also sought to make word-processing tools compatible with other languages and other scripts. Remington and other typewriter manufacturers knew that, after Roman script, Arabic script was the next most widely used script in the world, used for Arabic, Persian, Urdu, and—until the 1920s—Turkish.
Given these numbers, it is revealing how much of the history of Arabic word processing involves technicians faulting Arabic, and—even into the 1970s—calling for users to reform or change the script. Did they never wonder whether the technology, and its designers, might be the main issue? This chapter surveys twentieth-century efforts to redesign Arabic-script typewriter layouts to fit English-language typist norms. It then chronicles some of the attempts to frame the challenges of letter form and width, as well as key placement and type bar alignment, as problems that led to delays in rolling out Arabic-script word-processing programs, and as bugs that prevent toggling between Arabic and Roman scripts. This chapter argues that talk of such technical glitches and bugs obscures underlying hegemonic norms. The history of Arabic word-processing talk rehearses much older European-origin assumptions about the incompatibility not only of scripts but also of the social worlds that support them. As Mar Hicks has argued, these assumptions have the effect of making technological advances regressive rather than revolutionary, entrenching bias in new forms.1 As Halcyon Lawrence notes in this volume, similar assumptions about normative accents have reinforced voice technologies’ exclusion of nonnative and nonstandard speakers.2 Arabic is no more a conceptual problem for word-processing technologies than accented English is for voice technologies, and neither are bidirectional scripts—and nor are the people who daily operate in multiple languages and scripts.
Twitter’s ability to allow users to include multiple scripts in one tweet offers a technical fix—but the overall problem will not be resolved until we recognize that Euro-American cultural assumptions about unidirectional scripts and single-script fixity lie at its root. Once we address these assumptions, we can conceptualize technologies that fuel a global world in which many people live and work across multiple script forms.
In September 1873, New York State–based Remington Works—best known for its firearms—released its first commercial typewriter. Remington’s new product was a simple device that lacked a shift key and so printed every word in capital letters—although with a QWERTY keyboard that would remain the standard for any word-processing machine. It was promoted as a time-saving device for letter writing, with the added advantage of increased readability. However, because would-be typists first had to learn the QWERTY system, it was not until the emergence of typing schools and training courses in the late 1870s that the machine began to catch on in the United States—first as a business tool, and only decades later as a personal or leisure tool.3 Decades later, it finally delivered on its promise of “legibility, rapidity, ease, convenience, and economy,” in the words of the first typewriter catalog—effectively bringing the printing press revolution in reach of the typist (or at least the person or company she worked for).4
Accents were layered in by using “dead” keys, which type over the previous letter rather than moving to the next space. Non-Roman scripts—including Cyrillic, Greek, Hindi, and Arabic—presented more serious challenges: the typewriter’s keyboard layout, carriage return orientation, and kerning (spacing). “Of all the languages now written on the typewriter,” a 1923 industry history suggested, “the Arabic group presented the gravest mechanical difficulties”:
The Arabic character, as written, is not subject to any of the usual rules. It has in its own complete alphabet over one hundred individual characters; it writes backwards, i.e. from right to left; the characters are written on the line, above the line and below the line, and they are of various widths, requiring full spacing, half spacing, and no spacing at all. Here indeed was a medley of problems well calculated to tax ingenuity to the limit, and the Arabic typewriter is a crowning triumph of mechanical skill.5
With twenty-eight letters but multiple forms of each, the Arabic script, it argued, posed the largest and most difficult problem in typewriter design. Solving it was heralded as a singular triumph.6
That English became the norm was a historical accident, but it stuck. American typewriter designers privileged English language, not just Roman scripts, even when celebrating typewriter designs for other languages. For example, a 1903 New York Times article on the many type shuttle languages noted that an Arabic keyboard design had recently been patented and that a Japanese version was in process. Those designs, the article stated, featured the versatility of typewriters to tackle even the most challenging languages. But, it added, even Romance languages posed their own difficulties: “French, Spanish, and Scandinavian machines are like ours except that the keyboards contain certain accents that are not needed by us,” the article stated.7 Anything other than English, even if typed in Roman script, was a deviation from the norm.
The Ottoman citizen Selim Haddad obtained a United States patent for what is considered the first Arabic-ready typeface in 1899. A graduate of Beirut’s Syrian Protestant College, later known as the American University of Beirut, he had taught at a New York City–based night school for Syrian (Levantine) immigrants, run by his brother Ameen, that taught English to Arabic speakers.8 Starting in the late 1890s, United States consular officials around the Middle East had begun suggesting a need for Arabic typewriters, noting that “millions of people” in the region and elsewhere might appreciate the American import. In 1902, Gabriel Bie Ravndal wrote from Beirut: “There will be immediately a great demand for the Arabic typewriter, and before long, I take it for granted, several [companies] in the United States will be manufacturing and exporting Arabic machines.”9
In November 1899, Haddad filed Patent Number 637,109 with the United States Patent Office, containing a set of Arabic, Persian, and Ottoman Turkish types that could be installed on “any of the well-known typewriting machines, whether it be of the bar or lever description or of the wheel or segment or plate description.” His innovations started a lasting shift in Arabic type. “By the present alphabet the successive letters of a word are often connected in an oblique direction above or below the writing line,” he noted. “My letters are constructed so that all connect on a horizontal line and all bear a certain fixed relation to a base or writing line.” These changes economized space, permitting three typed lines in the vertical space usually taken by two handwritten lines. Haddad’s typeface also regularized the width of the letters, with a standard and double width, rather than individual variance among letters. His design thus fit Arabic within what Mullaney identifies as the key features of the QWERTY keyboard, with letters that all fell around a single baseline.10
But Haddad was unable to make Arabic letters isomorphic, with the same shape regardless of position, or to fit them all to the same amount of horizontal space. What he was able to do was to discipline the letters to minimize and regularize their variations in shape and position. He claimed that his type required only fifty-eight characters, thanks to his streamlining the positional variation of each letter.11 This remarkably simplified the 114 individual characters. In the Arabic script, as with other “cursive” or connected scripts, each letter has approximately four forms: a freestanding form (used when writing the alphabet), an initial form (used when the letter starts a word), a medial form (used when the letter is in the middle of a word), and a final form (used when the letter ends a word). There is no distinction between cursive and block print Arabic-script writing as there is, for example, in Roman-script writing. By Haddad’s count, the Arabic alphabet had twenty-nine letters or characters (including a connected lam/alif for “la”). His “new alphabet” streamlined these four forms into only two. Many critics would miss this feature, complaining that Arabic has too many letters, rather than having too many letter forms. Major United States newspapers covered Haddad’s invention as noteworthy news, including a 1902 article in the Washington Post.12 In 1904, a New York Times article noted that Haddad had prototype machines built as gifts for the Ottoman sultan and Egyptian khedive (ruler), but that commercial manufacture was still under development.13
Designing the type was only the first step: the machine itself, with Arabic-script type and a reversed carriage return, still needed to be built. Haddad’s typewriter layout addressed in theory the complexity of the Arabic script: the multiple forms of each letter, based on its position within a word. However, it was unable to address the bidirectionality of Arabic (and other right-to-left script languages like Hebrew): while Arabic words are written and read right to left, cardinal numbers are written and read left to right. Someone using this typewriter design would have needed to type out cardinal numbers in reverse order, starting with the ones or smallest digit rather than the largest.14 Haddad’s approach addressed both the typeface and how to lay it out on a standard typewriter, while other patent filers in the same period, like Arthur Guest and Ernest Richmond, addressed only the typeface. Guest and Richmond requested a patent for Arabic characters used in typewriters or printing, in which the characters had the same width at their base and all operated from one common point of alignment. “All the types made in accordance with this invention are on blocks of equal width [and] the character on each type commences from a point at the right-hand side, which point is the same in all the types and is herein referred to as the “point of [align]ment,” the patent specification explained. No effort was made to lay this typeface out on an actual typewriter; the inventors noted that “we do not limit ourselves to the precise forms of the characters illustrated in the examples shown, because they may be somewhat varied.” Instead, they sought only to standardize all Arabic letter base heights and widths and to fit a “standard typewriter.” But to actually type, one would have to replace the shuttle, wheel, and bar.15
Haddad’s innovations sparked early-twentieth-century efforts to restandardize the Arabic script, focusing primarily on the multiple forms of each letter and their varying widths. These script reforms did not explicitly address the font, but naskh became the standard font for Arabic typing on the typewriter, and later the computer, and nastaliq for Persian and Urdu—the equivalents of Courier and the similar-looking Remington fonts.16 For example, in 1922 Seyed Khalil was granted a patent for the keyboard layout that he had designed for the Underwood Typewriter Company. Initially submitted in 1917, this layout assumed that one form of each letter would be enough. “I have found that if I use forms of letters which are nearly perfect representations of the letters standing alone,” Khalil explained, “it becomes unnecessary to provide types for writing the modified connecting forms of these letters.”17 His layout provided terminal forms for those letters whose “terminal forms are very different from the medial forms,” but regularized their location so that terminal letters were all in the “uppercase” position and the nonterminal letters were all in the “lowercase” position of the keyboard, an arrangement he called “absolutely logical.” What logic justified reducing the script to a fourth of its original self? His design, he claimed, avoided changing existing typewriters, and also sped the typing process. His “cursive Arabic” script would bear only limited resemblance to Arabic, Persian, or Turkish handwritten texts. But it would produce a standardized script—and all it required was for hundreds of millions of people to relearn how to read. As with twenty-first-century voice technologies and non–native English speakers, typewriting technology expected users to adjust themselves to its limitations. In Lawrence’s words, these technologies are biased, and they discipline users for speaking with the “wrong” accent or using a script whose letters had too many forms.18
Kemal Ataturk ended Turkish’s script experimentation by romanizing the alphabet and script in 1928. The Latin alphabet was praised as supporting mass literacy, by making it easier for Turks to learn to read and write. It separated them from their rich Ottoman textual heritage, part of Ataturk’s efforts to give Turkey a national history that minimized the Ottoman role. In 1929, Remington shipped three thousand typewriters to fill an order from the Turkish government, for which it “had to construct new dies [the individual letters used in the typewriter], bringing the total cost to $400,000”; Remington’s foreign director explained the government wanted to “speed up its work by the use of these typewriters,” rather than writing documents and filling forms by hand.19 The New York Times described the typewriters as “aiding” Ataturk in his Westernization effort, and the “rat-a-tat-tat” of their keys represented the sound of modernity. (It also echoed the larger rat-a-tat-tat of the machine guns that contributed to the Ottoman Empire’s defeat in World War I and allowed France and Great Britain to crush the independence movements of the Ottoman Levant (Iraq, Lebanon, Palestine, and Syria).
Efforts to streamline the Arabic script continued into the mid-twentieth century. Mohammed Lakhdar, a Moroccan official who worked in the Ministry of Education, devised a script system incorporated into a typewriter in 1957, which the Ministry promoted as an aid to literacy. Arabic-script typewriters only included long vowels in their keys; they did not include short vowels, which are represented by diacritical marks, because designers felt that they would require too many keystroke combinations and complications. Yet because typed Arabic did not include vowelization, Lakhdar and others argued, only those with high literacy could read printed texts. Lakhdar’s system incorporated diacritical vowel markings but could do so on a standard typewriter layout, with no need for additional keys. “Machines with Latin characters can be easily utilized for Arabic with the Lakhdar system at the same cost,” supporters argued. The Lakhdar system promised to speed typing and reading further by requiring only one form for each Arabic letter.20
As an invention, typewriters are what historian Uri Kupferschmidt calls “small technologies” that overlap with consumer products or durables. “They were adopted, acquired, or appropriated as commodities on the [Middle Eastern] market, initially by elites, but gradually trickled down” toward mass consumption—like watches, sewing machines, gramophones, radios, umbrellas, or safety razors.21 It often took some time: these products had been developed elsewhere, and were only brought to Middle Eastern markets after having achieved mass-market success in the United States or Western Europe.22 Yet these small technologies found differing receptions. Unlike the sewing machine, which spread rapidly, typewriters penetrated the Middle Eastern business market slowly and shallowly. A lower adoption rate for some “small technologies” may have been due to the region’s smaller middle class, with less disposable income23 than in the United States. But in the case of the typewriter, Kupferschmidt suggests that lower literacy rates, the privileging of male scribal jobs, and the lag to a “pink collar” female stenographical job force, as well as the minimal private or leisure market, all negatively contributed to typewriter sales.24 Historical data regarding Middle East literacy rates is limited, but rates in the 1950s were significantly lower than those in Western Europe and North America.
But small technological inventions can have significant consequences. The typewriter shaped nontyped writing structure and script form: it set the standards, even though relatively few people owned or used a typewriter. This is most visible at the keyboard level: the typewriter layout envisioned first by Haddad has become the standard layout for Arabic computer keyboards, with a few small modifications (for example, the placement of the letter dal—Arabic’s “d” sound)—much as the QWERTY Latin-script keyboard transferred from typewriter to computer. The regularized letter widths and streamlined letter positions that characterized Khalil’s typewriter design have become standard for computer fonts—including Microsoft’s Traditional Arabic, which first appeared with Windows 2000.25
Arabic-script typing continued to develop into the later twentieth century. Seyed Khalil obtained another patent for Arabic-script typing in 1960, for example, which blamed “the many idiosyncrasies of the language” for inhibiting the development of “a practical Arabic typewriter . . . capable of operating at speeds and efficiencies of English typewriters.” He argued that the large number of Arabic characters in their various forms had forced the production of special typewriters with additional keys and space bars. He described Arabic’s connected script—“as in English handwriting”—as posing additional challenges, and repeated earlier assertions that the varying width of Arabic letters made typewritten interconnection even more difficult. As a result, he stated, aligning the type bar or bars required much more precision for Arabic-script machines. He critiqued existing typewriter designs—even those that streamlined Arabic letter forms—as failing to address the hyper-importance of type bar alignment in order to produce smoothly interconnected letters. Khalil’s new design still limited the Arabic letter forms to two per letter, while better aligning the type bar. Unlike the early-twentieth-century patents, Khalil’s described in meticulous detail each step in his typewriter mechanism design. Finally, the patent suggested that the new design could be applied equally well to a portable, standard, or electric typewriter—indicating that it would set the standard for future typewriter technology.26
As the century wore on, efforts to improve Arabic-script typewriting borrowed from the advances in early computing. For example: “A device developed by a Pakistani-born Canadian scientist promises to speed up typing Arabic script while preserving its aesthetic qualities,” noted a 1973 New Scientist article. “The pilot model . . . contains a logic circuit that permits an operator to use a standard 32-character western-style typewriter keyboard to print the 140 characters used on Arabic typewriters.” The machine, not the typist, calculated which letter form to use based on the letters immediately before and after it. As a result, it was supposed to make typing in Arabic faster. Typing a document in Arabic was described as taking “up to four times as long” as typing the same document in Roman script.27 Machines like this one were bridging the gap between traditional typewriters—manual or electric—and the emergent arena of computer-based word processing.
By the late twentieth century and early twenty-first century, the Arabic script was the third most commonly used, after Roman script and Chinese, and included those using Arabic script for Persian, Urdu, and other languages. Euro-American tech efforts shifted to developing electric typewriter and word-processing capacities for Arabic script, continuing the script modification and regularization approaches, but with less urgency and with no greater success.28 The electric typewriter offered only an stepwise improvement on the typewriter, rather than a wholesale innovation, although the 1973 machine’s “script processor” did point toward word processing and its emphasis on speed.
The personal computer, by contrast, rooted itself into the business and personal practices of the Middle East. Kupferschmidt describes the case as an instance of “technological leapfrogging,” defined as “the process whereby some developing countries can jump over several [developmental] stages to move rapidly from standard-modern to highly-modern technologies.”29 Historically, one key example is that of nineteenth-century France and Germany adopting British industrial technology like steam engines and railroad networks, which sped industrialization in those countries.30 European and North American businesses and private consumers were quick to adopt the personal computer, which facilitated its adoption in the Arabic-script world. The typewriter’s relatively low adoption also appears to have helped. Neither typewriters nor typing, or stenography as a profession, had developed deep roots in the region, as attested by the series of patents and reports on technological advances, which all describe Arabic typewriters as inefficient and difficult to use. Hence businesses and private consumers had less investment in typewriters and typewriting, making it possible for computers to play a positive rather than negative disruptive role. Few typists lost their livelihood due to the shift toward computers, for example.
Computers promised both powerful computational capabilities and potent word-processing capacities (at least in Roman-script languages). Word processing offered multifaceted corporate appeal, particularly with its on-screen editing capabilities, which streamlined document revisions. Like the typewriter, computer word-processing programs assumed a fixed writing orientation (left to right, at least initially), one default language, and one script form (Roman, at least initially). These assumptions of fixity all came from the typewriter, as well as United States–based developers’ experience working in monolingual professional environments—more specifically, in English and Roman script. Typewriter designers and computer programmers would share these assumptions with internet developers as well. Internet developers “were generally American, and were implicitly thinking only about how to facilitate communication in English.” ASCII, the standard and purportedly universal text-transmission protocol introduced in the 1960s, initially offered nothing for languages with accents and non-Roman scripts.31 Unlike the typewriter, computer word-processing programs promised instead many fonts and font sizes with a click of the mouse. (Some typewriters permitted formatting or emphasis shifts—italics and underlining, for example—but required the user to remove one “daisy wheel” and insert another when switching between italicized and regular text.)
An American University of Beirut research group described their efforts in the early 2000s to collaborate with MIT colleagues in developing a multilingual sustainability web platform that, in part, required the use of Lotus Notes, an IBM software platform for business that included email, calendar, contacts, discussion forums, file sharing, instant messaging, and database capabilities. The team reported that “the Arabic language version of this software was not fully developed” and had multiple flaws, stemming from the fact that Windows did not yet support Arabic fonts.32 In order to design fairly simple features like website buttons that Arabic-speaking users could click on for more information about a particular aspect of the initiative, the team laboriously pasted buttons and Arabic text into Word documents, imitating an Arabic graphical software interface. Even in the mid-2000s, Microsoft noted that its operations management suite “officially [did] not support . . . bidirectional, complex script languages, such as Arabic or Hebrew.”33 The same issues from a century before—a script running right to left, with numerals running left to right, and letters with multiple forms—remained stumbling blocks even for Microsoft’s robust server management suite.
The technical challenge, as described in a conceptual 1987 article on Arabic word processing, was twofold: doing the technical development work to create joinable letters and, as Becker put it, “intermixing scripts that run in opposite directions,” which he described as “a design puzzle.”34 Development efforts appear to have focused more on the first challenge than the second. For example, by 2001 Microsoft had developed a list of standard features for the Arabic script, derived from its OpenType font, Arabic Typesetting.35 The script features included not only the basic isolated, initial, medial, and final forms of each letter but also diacritical mark and alternative ligature (connecting) positions—suggesting a sophistication previously unseen. These efforts—to enable the use of Arabic-script typing rather than mixed-script typing, or the ability to move between multiple scripts in the same document—had mixed results. Users continued to report challenges using Arabic script in Office programs, and the Office approach to Arabic script was fairly well reflected in the title of its font: typesetting, which suggested its conceptual debt to typewriting and the printed word.
The issues with Arabic script in Microsoft-based word-processing applications have shifted over time but have not disappeared, particularly for Mac users. An English-language Apple discussions thread from January 2012, which began with novice user mka8b4, represents numerous similar discussions. “Just asking how to type Arabic letters in word, in more specific how to type letter or paragraph in Arabic language in Microsoft winword,” mka8b4 posted. “Your question refers to the generic problem of right-to-left (RTL) scripts,” replied a more advanced user. “Did you activate Arabic keyboard layout in System Preferences . . . ?” the responder continued. “If Word cannot handle RTL, you must choose something else; Nisus or Mellel [two word-processing systems that focus on multiscript and multidirectional functionality] are your best choices.” Another advanced user, identified as Tom Gewecke, added: “No version of MS Word for Mac has ever supported Arabic. You must use another app.” Picking up the thread again in August 2013, another user suggested that the issue was whether a particular font supported Arabic, but Gewecke argued that “MS itself agrees if you ask them that MS Word for Mac does not support Arabic (or any Indic script),” but that if the user could create an Arabic-script document in Word for Windows, they could open it successfully in Word for Mac.36 Five years later, a science-writing blog offered tips to manage the “bugs” that remained in Microsoft Word, suggesting that users add two additional toolbars in order to manage left-to-right and right-to-left formatting errors. The fix would “need to be done for each document in which you encounter this issue,” the writer cautioned, “but once you’ve done it, that style version should stay with the document even if you send it to someone else.”37
Critics of Word for Mac abound. “The issue is not that Apple doesn’t support Arabic language (it does),” noted Arabic–English translation firm Industry Arabic in January 2013. “The problem is that Microsoft Office for Mac is not built to work with right-to-left languages.” The firm suggested several work-arounds, including importing the document to Google Drive and editing it as a Google Doc, purchasing an Office clone like NeoOffice, or operating Windows through a virtualization program. “It remains frustrating that Mac users are forced to resort to these clunky work-arounds to work professionally in the world’s fifth most spoken language,” the firm concluded.38 Microsoft Office’s bidirectional capabilities have improved considerably since then, although it was not until March 2016 that Microsoft announced a security update that it said would allow Word, PowerPoint, and Excel to “fully suppor[t] Arabic and Hebrew.”39 Microsoft offers detailed instructions for changing various Word features for right-to-left languages—including column orientation and cursor movement.40 Yet users continued to experience difficulties, as this August 2017 query indicates:
When I’m working in English in MS Word for Mac (v 15.32), and switch to another application like a web browser in which I use Arabic, and then go back to Word, the document is no longer in English (I mean the language indicator on the bottom on the window). The keyboard layout is not Arabic, however, I cannot type left-to-right English in the doc. The only solution I’ve found is to save, close, and reopen the file, and even then I have to delete some text.41
What is being described here is a bidirectional language-enabled copy of Office for Mac, in which toggling between Word and another Arabic-enabled application confuses the orientation and encoding standards of the Word document. This in turn suggests that Office continues to have difficulty moving back and forth between uni- and bidirectional languages, and also that the bidirectional modifications remain unstable.
This is an important point. That the world’s dominant word-processing software has difficulty accommodating bidirectional and unidirectional scripts in the same document is a roadblock for the many people around the world whose life and work bridge script directions. It has produced notable gaffes in public signage, as the well-known Tumblr “Nope, Not Arabic,” which highlights Arabic word-processing errors, attests.42 The regular misunderstandings that arise between different linguistic cultures multiply when the very people who can work in both languages literally cannot type in both within the same document. This hurdle mandates awkward work-arounds. “Whenever I need to include Arabic in a presentation handout,” notes one scholar who gives presentations in English on Arabic and Hebrew materials, “I take a photograph of the text and insert it into a Word document as a jpeg.”43 Despite enormous complexities and advances in other fields, word processing still largely assumes that the word is not only linear—it is unidirectional.
Where you buy your computer matters. As Knut Vikør, professor of Islamic history at the University of Bergen, notes in his The Arabic Mac resource website, “On computers bought in the West, Arabic is normally just an extra language that is added”—just an add-on to “your basically English-European Mac.”44 And if it is not “added on” as part of the capability of each program used, it remains an invisible part of the computer. “Arabic capabilities do not reside in any particular program[;] they are generic to your computer itself.” Vikør blames cost-savings logics: “It has to do with saving money . . . [and] many companies cannot be bothered.” Those who might purchase a specialized program like Nisus Writer or use the free processors OpenOffice or Pages should recognize them as work-arounds, he notes, since sending a file created in those programs to someone using Word or another Arabic-incompatible word processor will likely strip the original document of its bidirectional functionality.
The 2017 release of new typefaces in Arabic (or, as in the Dubai font launched for Microsoft Office 365, simultaneously in Roman and Arabic script) redresses the early twentieth century framing of the “problems” of Arabic script, and a new belief in the positive power of font design. Sheikh Hamdan bin Mohammed bin Rashid al Maktoum, Crown Prince of Dubai, who spoke at the font’s launch, stated that the preciseness and distinctness of the design licensed “each human being” their “right to expression,” leading to an increase in “tolerance and happiness.”45 The Dubai font is claimed to be the first font “simultaneously designed in both Arabic and English” for general usage, with the capacity to operate in twenty-one additional languages.46 In other words, users must only toggle between keyboards in order to switch languages within the same document, rather than switching fonts; the line spacing is wide enough to accommodate both scripts without awkward intervals between Arabic and English text.47 Microsoft has set the font to install automatically in Office 2016’s base programs—Word, Outlook, Excel, and PowerPoint—for Windows and Mac users, and notes that it “works just as any other font in Office. Simply select the font from the font picker in your application to apply it.”48 Its clean, simple lines were intended to be legible and appropriate for multiple kinds of documents, as well as to privilege neither language in terms of aesthetics or design choices.49 It was intended as a showpiece, highlighting the ability of one font to serve two scripts, in two directions—and to enhance Dubai and the United Arab Emirates’ reputations on the global tech stage.50 And, in a way, it does. In 2017 a “solution” arrived. It is now possible to type bidirectionally in one document—thanks to this innovation from the Gulf.
The Dubai font casts new light on a 2003 special issue of the Journal of Computer-Mediated Communication. Editors Danet and Herring asked how people “communicating online in languages with different sounds and different writing systems adapt to ASCII environments.” Recognizing the then-dominant position of ASCII, they focused on users of non-English and non-Roman-script languages and their efforts to adapt to ASCII protocol, rather than vice versa. “What problems do they encounter,” they continued—a potentially technical question about communication glitches—“and what are the social, political, and economic consequences if they do (or do not) adapt?” ASCII’s difficulty with accented letters and with non-Roman scripts, they acknowledged, were technical problems that produced social, political, and economic consequences. Danet and Herring’s queries focused on internet communication rather than word processing. However, their final question highlights the limits of developing technical fixes for problems that stem from cultural assumptions. “Is there evidence of ‘typographic imperialism’?” they asked, referencing Robert Phillipson’s work on linguistic imperialism. Developing born-bidirectional fonts like the Dubai font may address some of the technical issues with bidirectional and unidirectional script switching. They do not address the social or political hegemonies undergirding the English-language, Roman-script dominance of computer programming, the internet, and computing infrastructure.51
Are born-multiscript fonts like Dubai the most appropriate or most likely locus of change for the future? Can they successfully address user difficulties in switching between languages and scripts, whether between applications, within applications, or within the same document? Or is it the operational platform that requires closer scrutiny—especially its creators’ underlying assumptions? Although the technical issues may appear resolved, the questions remain. Type design is too often considered primarily in aesthetic terms, and not enough for its support of underlying cultural logics of space, direction, orientation, and power. Type norms reflect our cultural assumptions; we need to address them in order to have more than piecemeal solutions to the challenges that people who live and work across multiple script forms face every time they fire up their laptops. As Hicks argues in this volume, powerful technologies often reinforce rather than relieve inequalities—to “preserve existing hierarchies and power structures.”52 American consular officials began promoting the typewriter in the Arab world during the imperial era, while word-processing software began circulating during the late Cold War; both technologies reflect their times and origins. It’s time to bring in more self-reflexive, less reductive ideas about script forms and script directions.
The Arabic script today is used in the official languages of twenty-six Arabic-speaking countries, as well as in a modified form in three Persian-speaking countries (Afghanistan, Iran, and Tajikistan) and two Urdu-speaking countries (India and Pakistan). It is additionally used to write other national or nationally recognized languages, including Uighur in China. While spoken and written in various communities elsewhere, recognition as an official language makes it more likely that government offices and businesses need to produce documents, and accept documents, produced in Arabic or modified Arabic script. Hence they are more likely to require word-processing and other document-production programs that support the use of Arabic script, either exclusively or in conjunction with other scripts or character languages. Individual users who wish to use Arabic-script languages for social purposes may be more likely to turn to transliteration or to communicate via social media, suggesting diverging needs for various Arabic-script users in the twenty-first century. They also provide a reminder that just as technological problems can have societal roots, they may also have societal solutions.
Several newer platforms that fall under the broad rubric of social media seem to operate with much greater fluidity in multiple scripts and languages. Twitter, for example, permits users to switch almost seamlessly between scripts within the same tweet.53 Facebook also supports multiscript communication. These newer platforms have evolved to offer users greater flexibility and higher functionality across multiple languages and scripts. What will it take for word-processing software to do the same? It might be that word processing, with its roots in typewriting and its founding assumptions about English, unidirectional script, and single-script fixity, is a dying fire, while more nimble platforms for human communication are igniting. They may spark a more agile concept of “document”—and the creative people who envision and develop that document may need to come from physical and conceptual locations beyond Western Europe and North America. Kavita Philip argues in this volume that we need “new narratives about the internet, if we are to extend its global infrastructure as well as diversify the knowledge that it hosts.” The same is true for the more foundational infrastructure that makes it possible for us to share knowledge by putting words on screens and pages.54 Asking better questions can help us blaze a trail away from the culturally laden assumptions that produced technology mired in unidirectional and monoscript norms toward the reality in which many people live and work across multiple script forms.
1. Mar Hicks, Programmed Inequality: How Britain Discarded Women Technologists and Lost Its Edge In Computing (Cambridge, MA: MIT Press, 2017).
2. See Halcyon Lawrence, “Siri Disciplines,” this volume.
3. The Story of the Typewriter, 1873–1923 (Herkimer, NY: Herkimer County Historical Society, 1923), 64–70.
4. Story of the Typewriter, 69.
5. Story of the Typewriter, 126–128.
6. The Story of the Typewriter stated that no functioning typewriters had yet been designed for ideographic languages like Chinese. See Thomas S. Mullaney, The Chinese Typewriter: A History (Cambridge, MA: MIT Press, 2017).
7. “Typewriters for Various Nations,” New York Times (January 18, 1903), 27.
8. “The Syrian Society of New York,” Stenographer 4, no. 1 (May 1893): 210.
9. Gabriel Bie Ravndal, Consul in Beirut (November 4, 1902), Consular Reports: Commerce, Manufactures, Etc. 71, nos. 268–271 (January, February, March, and April) (Washington: Government Printing Office, 1903).
10. Thomas S. Mullaney, “Typing Is Dead,” this volume.
11. Selim S. Haddad, Types for Type Writers or Printing Presses. US patent 637,109, filed August 12, 1899, issued November 14, 1899.
12. “Typewriter Prints Turkish,” Washington Post (November 10, 1908), 2.
13. “An Inventor from Mount Lebanon,” New York Times (August 21, 1904), SM8.
14. This point is often highlighted in Arabic textbooks. See, e.g., Sebastian Maisel, “Numbers,” Speed Up Your Arabic: Strategies to Avoid Common Errors (Oxford and New York: Routledge, 2015), 113.
15. Arthur Rhuvon Guest and Ernest Tatham Richmond, Types for Arabic Characters. US patent 639,379, issued December 19, 1899.
16. For a brief discussion of these stylistic or font standards, see Joseph D. Becker, “Arabic Word Processing,” Communications of the ACM 3, no. 7 (July 1987): 600–610.
17. Seyed Khalil, Typewriting Machine. US Patent No. 1,403,329, filed April 14, 1917, issued January 10, 1922.
18. Lawrence, “Siri Disciplines.”
19. “3000 Typewriters with Turkish Alphabet Shipped to Aid Kemal Westernize His Land,” New York Times (July 31, 1929), 27.
20. “New Typewriter for Arabs Shown,” New York Times (November 3, 1957), 121.
21. Uri Kupferschmidt, “On the Diffusion of ‘Small’ Western Technologies and Consumer Goods in the Middle East During the Era of the First Modern Globalization,” in A Global Middle East, ed. Liat Kozma, Cyrus Schayegh, and Avner Wishnitzer (IB Tauris, 2014), 229–260.
22. Kupferschmidt, “On the Diffusion of ‘Small’ Western Technologies,” 233.
23. See, for example, Relli Shechter, Smoking, Culture, and Economy in the Middle East (IB Tauris, 2006).
24. Kupferschmidt, “On the Diffusion of ‘Small’ Western Technologies,” 239.
26. Seyed Khalil, Typing Machines for Arabic Group Languages. US Patent 2,940,575, filed December 19, 1957, issued June 14, 1960.
27. “Electronics Make Typing Arabic Easier,” New Scientist 60, no. 827 (November 15, 1973): 483.
28. Regarding the quote in the section title, see Ramsey Nasser, “Unplain Text,” Increment 5 (April 2018), https://increment.com/programming-languages/unplain-text-primer-on-non-latin/.
29. Kupferschmidt, “On the Diffusion of ‘Small’ Western Technologies,” 239.
30. M. R. Bhagavan, “Technological Leapfrogging by Developing Countries,” in Globalization of Technology, ed. Prasada Reddy (Oxford: EOLSS Publishers, 2009), 48–49.
31. Brenda Danet and Susan C. Herring, “Introduction: the Multilingual Internet,” Journal of Computer-Mediated Communication 9, no. 1 (November 2003), https://doi-org.du.idm.oclc.org/10.1111/j.1083-6101.2003.tb00354.x.
32. Toufic Mezher, “GSSD-Arabic,” in Mapping Sustainability: Knowledge e-Networking and the Value Chain, ed. Nazli Choucri et al. (Dordrecht: Springs, 2007), 123–142.
33. Kerrie Meyler and Cameron Fuller, Microsoft Operations Manager 2005 Unleashed (Indianapolis: Pearson Education, 2007), 180.
34. Becker, “Arabic Word Processing,” 605–606.
35. See “Features for the Arabic script,” https://www.microsoft.com/typography/otfntdev/arabicot/features.htm, last updated 2002.
36. See “Just asking how to type Arabic letters in word,” https://discussions.apple.com/thread/3684500?start=0&tstart=0.
37. See Amanda W., “Fixing right-to-left text in Microsoft Word,” BioScience Writers (November 7, 2018), https://www.biosciencewriters.com/Fixing-right-to-left-text-in-Microsoft-Word.aspx.
38. See http://www.industryarabic.com/arabic-word-processing-guide-on-mac/, first published January 14, 2013. The page references a Facebook group and petition effort to convince Microsoft to develop an Arabic-language compatible Office for Mac, but as of August 2017, both had been removed from Facebook.
39. “MS 16-029: Description of the security update for Office 016 for Mac: March 16, 2016,” accessed August 28, 2017, https://support.microsoft.com/en-us/help/3138327/ms16-029-description-of-the-security-update-for-office-2016-for-mac-ma.
40. See, for example, “Type in a bi-directional language in Office 2016 for Mac,” accessed August 28, 2017, https://support.office.com/en-us/article/Type-in-a-bi-directional-language-in-Office-2016-for-Mac-d7bb1d52-4d82-4482-910f-d74b7c3bd468.
41. Private communication, August 28, 2017.
42. See https://nopenotarabic.tumblr.com/. In general, the submissions show disconnected (free-standing form) Arabic letters, often also reading left to right.
43. Private communication, September 11, 2017.
44. See “An Introduction to Writing Arabic on the Mac,” Knut Vikør, in The Arabic Macintosh, https://org.uib.no/smi/ksv/ArabicMac.html.
45. “Hamdan bin Muhammad Launches ‘Dubai Font’ into the World” [in Arabic], Al Bayan (May 1, 2017), https://www.albayan.ae/five-senses/mirrors/2017-05-01-1.2932379.
46. Dubai Font, “Hamdan bin Mohammed Introduces Dubai Font to the World and Directs Dubai Government Entities to Adopt It in Their Correspondence,” press release (April 30, 2017), accessed September 13, 2017, https://www.dubaifont.com/pressCenter/2-hamdan-bin-mohammed-introduces-dubai-font--to-the-world-and-directs-dubai-government-entities--to-adopt-it-in-their-correspondence.html. A critical piece by The National noted that the government of Abu Dhabi had in 2010 commissioned the first Arabic-English font, the Zayed; however, that font is proprietary and not available to the general public. See Nick Leech, “The Fine Print behind Dubai’s New Font,” The National (May 11, 2017), https://www.thenational.ae/uae/the-fine-print-behind-dubai-s-new-font-1.12697.
47. “How do I access the Latin or Arabic letters in the font?,” Dubai Font FAQ—Technical, accessed September 13, 2017, https://www.dubaifont.com/download.
48. “Using the Dubai Font in Microsoft Office,” Microsoft Office Support, accessed September 13, 2017, https://support.office.com/en-us/article/Using-the-Dubai-Font-in-Microsoft-Office-c862df16-ae0d-46d9-b117-aa3f41f9706e.
49. See “The Fine Print behind Dubai’s New Font.”
50. “Hamdan bin Muhammad.”
51. Danet and Herring, “Introduction.” See Robert Phillipson, Linguistic Imperialism (Oxford: Oxford University Press, 1992) and Linguistic Imperialism Continued (Hyderabad/New York: Orient Blackswan and Routledge, 2010).
52. See Mar Hicks, “Sexism Is a Feature, Not a Bug,” this volume.
54. See Kavita Philip, “The Internet Will Be Decolonized,” this volume.