Kol nebuvo „Palemono“: kompiuteriai, lietuviški rašmenys ir šriftai

PETRAS SKIRMANTAS Institute of the Lithuanian Language BEFORE THERE WAS PALEMONAS: COMPUTERS, LITHUANIAN CHARACTERS AND FONTS KEYWORDS: special characters, encoding, code page, ASCII, ANSI, Unicode, text accentuation, phonetic transcription. Discussion about Lithuanian computer fonts is continually renewed: efforts are made to determine how many there are, what kinds, and for whom they are intended, in which cases they are needed, how widely they are used, and so on. Recently, another opportunity to discuss this arose when the State Commission of the Lithuanian Language and the Institute of the Lithuanian Language held a discussion seminar, Palemonas and Other Lithuanian Fonts: Use and Prospects. This article also attempts to touch on what was discussed there. Practical needs prompted us to take up making computer fonts at home—or, more accurately, modifying them—as the saying goes, necessity is the mother of invention... In the first half of the last decade of the previous millennium, after 1990, computers of the time began to appear and grow increasingly common even in the circles of philologists and Lithuanian-language specialists. First, they were becoming increasingly affordable; second, many were brought or donated as aid from abroad, since people overseas acquired newer, more modern ones and gave away or sold relatively cheaply their outdated machines to us. Computers then were quite unlike those of today, with far more modest capabilities. But since even computers of that time could, among other things, be used to write—as we began to say then, to type text—they very quickly displaced typewriters. The convenience this brought was enormous and decisive: when writing or rewriting text on a computer, a concept such as draft became meaningless, because all corrections of errors or editorial changes replaced the relevant passages in the text, and both on screen and in print only a clean copy was visible at all times. Soon, for philologists, as for many people doing other kinds of work, writing and typing text became the main task performed on a computer. Since computers reached us in various ways and were often far from new, the problem of being able to write correctly in Lithuanian on them arose right from the start: we wanted and needed all the characters used in our current orthography to be visible both on screen and in print BENDRINĖ KALBA 92 (2019) www.bendrinekalba.lt ISSN 2351-7204 PETRAS SKIRMANTAS. Kol nebuvo Palemono: kompiuteriai, lietuviški rašmenys ir šriftai | 2 doi.org/10.35321/bkalba.2019.92.09 characters, first and foremost all the letters of the modern Lithuanian alphabet (including ą, ę, į, ų, č, š, ž, ė, ū), as well as other writing signs found in texts. This was needed not only by philologists and Lithuanian-language specialists, but by anyone who wanted to write or edit Lithuanian texts in their field on a computer. And the means to do so were not always available... It is worth digressing briefly to recall that, even back then, computers came into the hands of philologists in two main kinds, and they differed considerably from one another both in capabilities and in ease of use (their modes of operation and, more generally, the nature of “communication” with the user, the so-called interface). Some were designed and manufactured by the Apple company, others by International Business Machines (IBM). Around the mid-1980s, the Apple company began producing Macintosh computers (named after one variety of apple), which later became better known as Mac. From the outset, they had a special control device—the mouse—and a graphical interface with the user (graphic user interface, GUI), which arranged almost everything needed for working on a computer into carefully structured lists displayed on screen in graphical windows specially drawn for that purpose. It was enough to point the mouse at the required item in a list, and the computer “understood” what it was supposed to do. The windows also showed the result of an action—the text or graphic document being processed and its state after one action or another. The basic principles of the graphical interface spread widely and remain in force to this day. At the same time, similar PC-series computers (PC being an abbreviation of the phrase Personal Computer) began to be manufactured by IBM, and for a while they had no graphical interface. All their operating commands had to be entered (typed) on the keyboard, and the interface with the user was purely textual: both the commands entered and the results of the computer’s actions were displayed on screen as text made up of identical, standard-sized and standard-shaped characters confined to character cells. A character cell is a standard rectangle consisting of a certain number of dots arranged in uniform columns and rows. The keyboard and the screen divided into character cells together are called a text terminal; for a time, this was the main means of exchanging information with an IBM PC-type computer. Working on such a computer was therefore somewhat cumbersome: one had to remember numerous operating commands and enter them and their required parameters (data, references) correctly, without mistakes. One could say that a person corresponded with such a computer: the person entered commands and data on screen in standard text, while the computer displayed the results of executing those commands. In some operating systems, an interface based on the text-terminal principle remains the main means of exchanging information with a computer even today, and is especially favored by professionals. Most people, however, prefer the graphical interface mentioned earlier, considering it more convenient. It is therefore natural that it was gradually developed for IBM PC-type computers as well; it was called BENDRINĖ KALBA 92 (2019) www.bendrinekalba.lt ISSN 2351-7204 PETRAS SKIRMANTAS. Kol nebuvo Palemono: kompiuteriai, lietuviški rašmenys ir šriftai | 3 doi.org/10.35321/bkalba.2019.92.09 an apt name Windows, directly pointing to the main interface element—graphical windows. By the mid-1990s, Windows had grown into an independent operating system, and from then on the graphical interface became the principal way people interacted with IBM PC-type computers. The operation of both types of computers—those Apple-made Macintosh computers and IBM PCs—and, more generally, working with both types of computers became quite similar, although essential “internal” differences remain. Among philologists, as in many parts of Lithuania, IBM PC-type computers became more widespread. They were more accessible, cheaper—one might say, more democratic—and it was easy to install software of various origins on them, especially the necessary application programs, including even home-made ones, created by the users themselves or commissioned from programmer acquaintances. Alongside the relatively rare original PC-type computers manufactured by IBM, counterparts made in Southeast Asian countries—or even assembled here in Lithuania from components manufactured there—began to spread rapidly. These came to be called computers compatible with IBM PCs... It is worth noting that essentially no similar Macintosh imitations existed. Philologists who mostly used IBM PCs and computers compatible with them rarely encountered major problems with correct, standard Lithuanian orthography. Even in the PC-type computers of that time, running the DOS operating system (English Disk Operating System), with which information was exchanged through a text terminal, fairly well-defined schemes were already in use for encoding written characters displayed in screen positions, known as code pages or code tables (English code page, CP). There were several of them, and they provided codes not only for the letters of the English alphabet but also for characters used in the writing systems of some other countries. Computer fonts (English fonts) were distributed along with the operating system—software components integrated into it that enable characters to be represented as visible graphic images, perceptible to the human eye. There are two main devices capable of doing this: the screen and the printer. The more problematic device, requiring more ingenious adaptation to the computer, was the printer of that time: much depended on its operating method, type, manufacturer, and so on. But it is important to emphasize that both devices perform their task only under the computer’s control, and the computer obtains the essential information it needs from computer fonts. Naturally, correct Lithuanian orthography was not the main concern of Microsoft, which developed software for IBM PC-compatible computers (and continues to develop it for many computers today). On the other hand, however, the basis of our present-day orthography—the Lithuanian alphabet—is neither original nor particularly unique: most of its letters are shared, used in their writing systems not only by Lithuanians but also by other peoples. Truly Lithuanian letters, not used by anyone else, at least BENDRINĖ KALBA 92 (2019) www.bendrinekalba.lt ISSN 2351-7204 PETRAS SKIRMANTAS. Kol nebuvo Palemono: kompiuteriai, lietuviški rašmenys ir šriftai | 4 doi.org/10.35321/bkalba.2019.92.09 anywhere else in Europe in practice, are probably limited to ė, į and ų. Poles use ą and ę in their writing system; Czechs use ć, ś and ź; and Latvians use ü. Nevertheless, nine letters—ą, ę, į, ų, č, š, ž, ū, ė—are often set apart from the other letters of the Lithuanian alphabet. And rightly so: they are all double, two-part letters, composed of two components—a base letter (a, e, i, u, c, s, z) and an auxiliary mark added above or below it: a ogonek, caron, macron, or dot. All the other 23 letters of our alphabet—a, b, c, d, e, f, g, h, i, y, j, k, l, m, n, o, p, r, s, t, u, v, z—are single-part letters, consisting of only one component (the only possible exceptions to this claim might seem to be lowercase i and j, which have a dot above, but it should be remembered that uppercase I and J have no dot, whereas both ė and Ė do, so these pairs of letters are different). It should also be noted that the single-part letters are much more “international” than the two-part ones: they are used in the writing systems of many European countries, and all 23 are part of the English alphabet. It is especially important that these two groups of letters fell into different code groups as schemes for encoding written characters in computers were being developed. The single-part letters of our alphabet, also being letters of the English alphabet, were assigned codes first, and codes that later ceased to change and have remained constant to this day. They belong to the so-called standard character set (English standard character set), governed by the old ASCII standard (English American Standard Code for Information Interchange), which emerged in the 1960s. Under this standard, integers from 0 to 127 are used to encode characters; written characters (letters, digits, punctuation marks) use codes from 32 to 126. This encoding scheme (sometimes simply called an encoding) was established before “small” computers appeared. It was intended for various telecommunications devices, and computers simply adopted and appropriated it as they emerged. However, as computers grew in number and spread, the characters in the standard (ASCII) set began to prove insufficient, since it did not contain all the characters needed to write texts not only in English but also in other languages. To meet the growing demands of computer-based writing, more integers—from 0 to 255—began to be used for character encoding, as the internal architecture of computers allowed this. The first 128 codes (from 0 to 127), governed by ASCII, remained unchanged, but 128 additional codes (numbers from 128 to 255) were introduced to encode other characters not included in the standard set. English speakers often describe this expanded set of codes (code space) as Extended ASCII, but this is not an addition to or extension of ASCII (the standard or the encoding scheme it governs). Rather, it is an expansion of the code set so that it encompasses the unchanged ASCII encoding scheme while also providing 128 additional codes for other required characters whose encoding is not governed by ASCII. This gave rise to a 256-code allocation model, dividing the entire code set into two equally sized subsets—the fixed subset, governed by ASCII and encompassing codes from 0 to 127, and the free subset, encompassing codes from 128 to BENDRINĖ KALBA 92 (2019) www.bendrinekalba.lt ISSN 2351-7204 PETRAS SKIRMANTAS. Kol nebuvo Palemono: kompiuteriai, lietuviški rašmenys ir šriftai | 5 doi.org/10.35321/bkalba.2019.92.09 255. This model is precisely what the English description Extended ASCII denotes. Its established free subset of codes made it possible for numerous different encoding schemes to emerge on the basis of the model, called, as already mentioned, code pages or code tables and designated by a wide variety of names. Their fixed subset, the constant part, was the same in all of them and coincided with ASCII, but the assignment of codes in the free part to characters varied considerably, depending on the character set for which the code page was created. As the variety of code pages continued to grow, some began to appear whose free portion also encoded the compound (double) letters of the Lithuanian alphabet, more often only those needed not just for our orthography. Code pages that provided codes for all the letters of the Lithuanian alphabet also appeared; our computer specialists contributed to this as well. Computer fonts corresponding to those code pages also emerged, allowing the written characters encoded with the assigned codes to be used in practice on computers. There was also some confusion: in the first half of the final decade, there were already several code pages for the DOS operating system containing all the letters of the Lithuanian alphabet—their numbers ranged from 770 to 774, followed by 775, 776, 777, and 778. The codes for Lithuanian letters coincided in some of these code pages and differed in others. As the number of computer fonts displaying the characters encoded in those code pages increased, one had to be discerning and cautious. It was also necessary to map the varying codes for the nine double letters of our alphabet, often called distinctive or specifically Lithuanian letters, to the keyboard and its corresponding keys (for understandable reasons, mapping the codes for single letters included in the standard ASCII character set to the keyboard had never posed any problems). On the other hand, however, all this made it possible to write correctly in Lithuanian even on computers running the DOS operating system. The graphical interface Windows, designed for IBM PC-type computers, which soon evolved from a DOS add-on into a standalone operating system (the first to do so was Windows 95), fundamentally changed the display of characters on screen, computer fonts, and the overall state of computer-based writing. Character-cell graphics disappeared, and different, new kinds of fonts took hold, called vector fonts, which create the visible forms of characters using vector computer graphics—drawing character outlines with Bézier curves. Inkjet and laser printers, capable of reproducing those outlines on paper with great precision and in the finest graphic detail, became more common. As a result, texts and documents prepared on a computer and printed on such a printer came to resemble documents produced by printing presses. Many new kinds of fonts spread along with the Windows graphical environment and were an integral part of it, but it was also possible, when needed, to supplement the “system” font set with other fonts purchased separately. BENDRINĖ KALBA 92 (2019) www.bendrinekalba.lt ISSN 2351-7204 PETRAS SKIRMANTAS. Kol nebuvo Palemono: kompiuteriai, lietuviški rašmenys ir šriftai | 6 doi.org/10.35321/bkalba.2019.92.09 Character encoding also began to stabilize. Despite the variety of local code pages, universal international encoding standards became increasingly important. Two are particularly notable: ISO/IEC 8859, comprising a dozen or so parts developed gradually, and ISO/IEC 10646. The former specifies how the characters needed by most European writing systems are to be encoded in code pages containing 256 codes each, while the latter greatly expanded the set of integers that can serve as character codes—it now exceeds two billion! Most of this set remains unused; so far, only nearly 140,000 characters have been encoded using its codes... Another character-encoding standard (not an international one) emerged in parallel—Unicode. It, too, was intended to expand the set of character codes. Its outlines were prepared in 1988; the Unicode Consortium was established in early 1991, and in the autumn the first volume of the Unicode standard was published, according to which 65,536 integers—from 0 to 65,535—could serve as character codes. Soon, the developers of the ISO/IEC 10646 and Unicode standards agreed to work in coordination; thus, the two standards were (and still are) harmonized and synchronized, and their character encoding has been entirely identical ever since, although the purpose and functions of each standard have remained distinct. For various reasons and because of certain requirements, the total number of codes that can actually be used, usually called the code space, has shrunk considerably and currently stands at just over 1.1 million; of these, as noted, codes have already been assigned to about 140,000 characters, or roughly 12.5 percent. The union of the two standards is generally called Unicode. The name is apt, referring to efforts to unify the encoding of all kinds of characters used in computers, making it uniform throughout the world. The aforementioned character-encoding standards are universal and oriented toward the needs of information technology—they were not created solely for IBM PC-type computers or their graphical interface. However, when computer fonts are developed for various types and purposes of computers and for different operating systems, efforts are made to adhere as closely as possible to the character encoding specified by these standards, because this is useful: unified character encoding ensures that written information conveyed through written characters will be transmitted, received, and reproduced accurately, regardless of the hardware or software used. The software manufacturer Microsoft, which created, developed, and distributed the graphical interface Windows—which evolved so rapidly that, after a while, figuratively speaking, it swallowed the DOS operating system and itself became an operating system—used its own code pages, each covering 256 character codes, for the vector fonts distributed with it. They were given distinctive names, but differed from one another and from DOS code pages only in the free subset of codes, that is, in the assignment of codes from 128 to 255 to characters. The fixed subset, governed by ASCII, was also left unchanged (incidentally, it was unchanged in all parts of ISO/IEC 8859 and in Unicode as well). With the first versions of Windows BENDRINĖ KALBA 92 (2019) www.bendrinekalba.lt ISSN 2351-7204 PETRAS SKIRMANTAS. Kol nebuvo Palemono: kompiuteriai, lietuviški rašmenys ir šriftai | 7 doi.org/10.35321/bkalba.2019.92.09 a code page numbered 1252 began to be used (code page 1252). Later it was also referred to in abbreviated form as CP-1252, as Windows-1252, and simply as Latin 1. More code pages were subsequently added, along with more characters encoded by them, based on both the Latin and other alphabets. To emphasize that these were code pages for the Windows environment, rather than DOS, they were sometimes called ANSI code pages, since the original version of the first one, CP-1252, was based on the initial draft of a code page developed by the American National Standards Institute (American National Standards Institute, ANSI). That draft later evolved into the ISO/IEC 8859-1 code page, but the name ANSI nevertheless became established in Microsoft’s working terminology and became a collective name for Windows code pages. It should be added that code pages for the DOS operating system also had a collective name—OEM. As has been noted more than once, code pages are character-encoding schemes that specify which code—an integer no greater than 255—is to be assigned to each character. A code—the character’s digital identifier in a computer system, a conventional symbol by which it is recognized—is much like a radio station’s call sign or a Lithuanian citizen’s personal identification number. In English-language computing literature, a code is now commonly called by the apt, well-formed word codepoint (sometimes written code point), indicating that it is a point in a set of codes, one of the numbers in that set. Nevertheless, it should be emphasized that a code is not some immanent attribute of a character, hidden “inside” it—it is freely selected and assigned to the character on the basis of a certain convention. Therefore, the same character, if it occurs in different code pages and/or different encoding standards, can in principle receive different codes, although this is undesirable and should be avoided. Codes are needed not only to encode characters in computer fonts; above all, they are needed so that information expressed in written characters can circulate in a coordinated way within a computer system, and so that the various software and hardware components that send and receive information preserve it unchanged, without distorting it or turning it into a stream of random signals—that is, informational noise, or entropy. As computers improved and it became possible to use integers many times greater than 255 as codes, the importance of code pages diminished somewhat. In Unicode, the standard that greatly expanded the code space, the concept of a code page is no longer used; instead, different groups of characters are distinguished according to their purpose and usage, and are called scripts, while subdivisions of the code space are more often called blocks. However, Microsoft consistently adheres to the principle known as backward compatibility (sometimes called historical compatibility): the aim is to ensure that older programs used in the past can also run on computers controlled by new versions of modern Windows. For this reason, current versions of Windows support a dual character BENDRINĖ KALBA 92 (2019) www.bendrinekalba.lt ISSN 2351-7204 PETRAS SKIRMANTAS. Kol nebuvo Palemono: kompiuteriai, lietuviški rašmenys ir šriftai | 8 doi.org/10.35321/bkalba.2019.92.09 encoding: both the innovative kind, compatible with Unicode, and the old-fashioned kind, conforming to ANSI code pages created earlier. This circulation of information inside the computer is controlled by a set of software functions in system libraries Windows API (API—from English Application Programming Interface), and many of its functions for working with characters and their sequences exist in two versions: A, for characters encoded with ANSI codes, and W (English wide), for characters and their sequences represented by Unicode codes. Since the correspondence between the ANSI code assigned to a character in the various code pages and the code assigned to it by Unicode is one-to-one, it is easy to convert one kind of code into another “in both directions”—one only needs to know which ANSI code page to use for the conversion. Characters are recoded into ANSI codes when necessary because of certain internal system features and because of legacy applications that still exist and do not recognize codes greater than 255. Modern text applications generally manipulate characters encoded using long (English wide) codes. Unicode character encoding is now the norm. As vector fonts took hold, correct Lithuanian spelling was taken care of. This, too, is a contribution by our computer specialists. The period when computers, displacing typewriters, began to appear in ever greater numbers on our philologists’ desks roughly coincided with the establishment of the first versions of the graphical interface Windows. Admittedly, that interface could function properly and usefully only on newer computers with at least an Intel 80386 processor and a VGA-standard video adapter (graphics card) and monitor, but by no means did all the computers philologists were buying at the time meet these requirements. As Windows versions changed, requirements for computer hardware kept increasing—or, more precisely, the reverse: as hardware capabilities grew, the conditions emerged for new, improved versions of the graphical interface Windows to appear. At the time in question, version Windows 3.1 was already beginning to spread, followed somewhat later by Windows 3.11, and vector fonts containing all Lithuanian letters were already available. First to mention here is the font TimesLT, prepared by Fotonijos, which became very widespread and was used by Lithuanians for a long time. Its letterforms imitated the extremely popular typeface Times. Soon there appeared equivalents of TimesLT, too, with letterforms imitating other typefaces, but encoding Lithuanian letters in the same way as TimesLT. Characters in these fonts were now encoded according to a common code page then being developed by computer specialists in Lithuania and Latvia, with the grandiose name BalticRIM—supposedly “Baltic rim” or “Baltic ring”... It was certainly also nice that the name unexpectedly reflected the abbreviation of the Ministry of Communications and Informatics, which was then in operation and overseeing information technology matters. The ANSI section of that code page gathered characters needed for the writing systems of most countries around the Baltic Sea, but not included in its ASCII section. Of course, among them were our nine distinctive, doubled letters. And beginning with the Windows-95 PE version (PE—from BENDRINĖ KALBA 92 (2019) www.bendrinekalba.lt ISSN 2351-7204 PETRAS SKIRMANTAS. Kol nebuvo Palemono: kompiuteriai, lietuviški rašmenys ir šriftai | 9 doi.org/10.35321/bkalba.2019.92.09 English Pan European), Microsoft effectively established this code page as the one intended for the Baltic countries, Windows-1257 (called Windows Baltic, CP1257, CP1257 BalticRIM, and so on). In it, Lithuanian letters and all the characters of the Baltic countries are encoded with the same codes that had already become established in Lithuania with TimesLT and other fonts using that encoding. Thus, a character set evocative of the name Baltijos žiedas (“Baltic ring”) took shape and became established, finding its way into two more code pages: 775, intended for the DOS operating system still in use at the time, and, later, the thirteenth part of the international standard ISO/IEC 8859, published in 1998 (ISO/IEC 8859-13:1998 Information technology – 8-bit single-byte coded graphic character sets – Part 13: Latin alphabet No. 7), which is also often unofficially called BalticRIM, Baltic RIM or Baltic Rim. The character set is the same, but the encoding—apart from the codes for quotation marks and the apostrophe—essentially matches only between Windows-1257 and ISO/IEC 8859-13; encoding the characters with OEM code page 775 differs considerably. Vector computer fonts fundamentally changed the appearance of text documents produced on computers. They began to differ considerably from documents typed on typewriters, or typescripts, and increasingly resemble printed matter produced by printing presses using typographic methods. The use of vector computer graphics to create character forms made it possible to change letter size (point size) across a wide range as needed, intended, or desired, without deforming their outlines at all; it also became possible, given a suitable set of vector fonts, to change the weight of letters of any size (the width of the strokes that form them), set selected letters in italics, and adjust spacing between letters and lines. People liked and appreciated the expressive, typographic appearance of documents prepared on computers, and so did philologists; consequently, almost immediately, one of the components of the Microsoft Office application suite distributed by Microsoft became—and remains to this day—one of their principal tools for writing on computers: the word-processing program Microsoft Word, since it allows the aforementioned capabilities of vector fonts to be put into practice relatively easily. Even today, there are philologists who cannot be persuaded that texts can be written using programs other than Word, and that in some cases it would be more practical to prepare them using other software tools. The clinching argument is: “It comes out prettiest in Word...” The Lithuanian letters available in computer fonts, and thus the ability to write Lithuanian text correctly, without violating current spelling norms, met—and continue to meet—the practical needs of many institutions that prepare written documents and of many people for whom writing is an integral, and often primary, part of their professional activity. For a writer, journalist, literary critic, screenwriter, or someone in a similar profession, these capabilities are often sufficient; little more is needed. Philologists, however, especially linguists, use in their written work not only the familiar letters of the Lithuanian alphabet, but also many other kinds of signs and a great variety of characters, sometimes BENDRINĖ KALBA 92 (2019) www.bendrinekalba.lt ISSN 2351-7204 PETRAS SKIRMANTAS. Kol nebuvo Palemono: kompiuteriai, lietuviški rašmenys ir šriftai | 10 doi.org/10.35321/bkalba.2019.92.09 even highly specific and unusual ones. Understandably, the typewriters used earlier could not print these special characters, so it was customary to write them by hand into the typescript of a philological text. A typescript was usually (with the possible exception of dissertations and the occasional other kind of work retyped on a typewriter) only an intermediate version of the text, something of a working aid, while the text acquired its final graphic form in the printing house, where it was reset in its entirety using metal type. The variety of type available there was far greater than the range of characters covered by a typewriter. Usually, the printing house could find metal type corresponding to the special characters written in by hand; if it could not, it would either make them separately to order or ask the author to replace a character with another one it had available. When moving from typewriters to computers, philologists saw that computers could insert far more and much more varied characters into a text than a typewriter could, and began to want to use every character they needed in their texts—every conceivable one. Fulfilling such wishes was neither impossible nor very difficult: it was enough to insert, by some means, even by teaching oneself, additional characters absent from the font of the required typeface (most often from the Times family). Ways and means of doing this were found. And once it had been practically demonstrated that this was possible—that, in principle, the character set on a computer could be expanded and augmented at will—the number of wishes and desired special characters began to grow almost exponentially... Thus began the somewhat peculiar practice of modifying computer fonts, and it lasted for many years. The special characters required for philologists’ work, especially in various fields of linguistics, can be divided into several groups. The following are worth mentioning: 1) Lithuanian letters marked with accent signs. These are needed, incidentally, not only for scholarly works by philologists, but also for writing in other fields—for textbooks and other teaching materials, various dictionaries, reference works, and so on. Counting uppercase and lowercase letters separately, there are 68 Lithuanian letters marked with accent signs in all. None occurs in the standard Lithuanian alphabet, though some are found in the alphabets of other countries. 2) lietuvių kalbininkų vartoti ir vartojami fonetinės transkripcijos ženklai. Lietuvių kalbininkai tarmei kalbėsenai užrašyti dažniausiai naudoja ne tarptautinės fonetinės abėcėlės (TFA, angl. International Phonetic Alphabet, IPA) ženklus, bet savitą fonetinę rašybą, kurios ištakos – vadinamoji kopenhaginė transkripcija ir kurią lietuvių tarmėms pritaikė Jurgis Gerulis (vok. Georg Gerullis), o vėliau detalizavo ir papildė Aleksas Girdenis. 3) special characters needed for publishing written heritage. Old Lithuanian writings contain many such characters that did not make it into the modern Lithuanian alphabet. But when citing BENDRINĖ KALBA 92 (2019) www.bendrinekalba.lt ISSN 2351-7204 PETRAS SKIRMANTAS. Kol nebuvo Palemono: kompiuteriai, lietuviški rašmenys ir šriftai | 11 doi.org/10.35321/bkalba.2019.92.09 written heritage, it is customary, whenever technically possible, to reproduce those characters as they appear in the original. Some linguists, for example Professor Vincas Urbutis, took a rather scrupulous approach to this. 4) special characters needed for work in historical grammar, Baltic studies, and Indo-European studies. These are most often characters from various alphabets (Latin, Greek, Cyrillic), supplemented with various specific auxiliary signs. The ultimate goal of philologists’ work was, and for the most part still is, a printed form: a work is considered fully complete when it becomes a book or an article in some publication. As long as printing houses produced printed works in the old way—resetting text in metal type, then breaking it into pages (laying it out) and, from pages arranged in the required order and imposed into signatures, casting printing plates—the text intended for the printing house, even when prepared and printed on a computer, had essentially the same status as a typescript had earlier. Thus, special characters printed into it rather than written in by hand were still mostly a matter of appearance or of a somewhat idiosyncratically understood prestige; the printing house would reset them in its own type in any case. In other words, special characters could still be added by hand—the printed work did not reveal this. But around the middle of the last decade, the situation began to change fundamentally: as improved fonts and laser printers made it possible to print pages whose appearance came close to that of traditional printing, printing houses began to abandon the “metal” resetting of printed works and to make printing plates directly from computer-printed pages. At this turning point, page layout became the concern not of printing houses but of publishers and those preparing publications; printing houses wanted to receive finished, printed pages. At first, these had to be printed on special film (transparencies); later it was enough to print to a PDF file using a virtual printer (a specialized program). In principle, a missing special character or one of its components—for example, a mark indicating pitch accent—could still be written by hand on a page transparency, but this was no longer possible in a PDF file. Characters written in or supplemented by hand would then appear in the print run, be visible in every copy, and, standing out from the other characters, spoil the overall appearance of the printed work. Thus, having all characters, including special ones, on a computer—which had previously been more of a luxury—became a necessity. Without them, it was no longer possible to prepare high-quality printed works. The preparation of printed works that did not require special characters was less affected by such changes in printing technology. Publishing houses simply acquired additional work in layout, along with people to do it, and technical equipment that made it possible to print high-quality transparencies of prepared layouts. But those preparing linguistic publications that required many and varied special characters, and BENDRINĖ KALBA 92 (2019) www.bendrinekalba.lt ISSN 2351-7204 PETRAS SKIRMANTAS. Kol nebuvo Palemono: kompiuteriai, lietuviški rašmenys ir šriftai | 12 doi.org/10.35321/bkalba.2019.92.09 The break affected publishers quite painfully. Their computers usually lacked many of those characters, and it was unclear how many and which ones might be needed. As a result, some of them began to lean toward the idea that, like printing houses, they would be better off receiving finished, typeset linguistic publications—in effect, computer-laid-out versions of them. Those that undertook the layout themselves managed as best they could. For these reasons, the Faculty of Philology at Vilnius University began attending not only to the scholarly but also to the technical preparation and layout of the linguistics journals Kalbotyra and Baltistika. They also began considering the technical preparation for print of other publications, including some books on linguistics. Thus, without specialized linguistic characters, not a step could be taken... Work began with Lithuanian letters bearing stress marks. In fact, the shortage of these had already been felt earlier by the company Fotonija, which was developing various tools to support Lithuanian writing on computers and had begun compiling the first computerized dictionaries. It modified some DOS code pages to include Lithuanian letters, inserting into them stress-marked Lithuanian letters and the phonetic transcription symbols needed for dictionaries (this gave rise to code pages 776, 777, and 778). The Fotonijos font TimesLT variant with stress-marked Lithuanian letters TimesLTAccented also became very popular and was used by many for a long time; it too, naturally, included the transcription symbols needed for dictionaries. The unaccented letters of the Lithuanian alphabet in the TimesLTAccented font retained the same codes as in TimesLT, that is, the Windows-1257 code-page codes, whereas the stress-marked letters were placed in the ANSI portion in place of other non-alphabetic characters, thereby “appropriating” their codes. Consequently, Lithuanian text with stress marks prepared in the TimesLTAccented font differed considerably from unaccented text: all stressed letters received different codes from their unstressed counterparts, and neither spell-checking nor hyphenation software could handle such text. This was an obvious obstacle to publishing linguistic texts on computers. So people began to consider a different way of preparing texts with stress marks. The idea, in fact, came from MS Word itself, where bold (English bold) and italic (English italic) passages are produced by changing fonts. The character codes in those fonts do not change, but the appearance of the characters—their graphic representation—does. This does not interfere at all with spell-checking or automatic hyphenation. Thus arose the idea that changing the font could make a stressed letter replace an unstressed one, and vice versa. The idea was put into practice: a modification of the Times typeface font, LT-Times, was created, comprising all four stylistic variants—regular upright, regular italic, bold upright, and bold italic. The letters were encoded according to the Windows-1257 code page, so texts set in the TimesLT font also displayed correctly. A couple of additional characters were assigned the undefined ANSI codes of Windows-1257. Most importantly, however, alongside that four-member LT-Times BENDRINĖ KALBA 92 (2019) www.bendrinekalba.lt ISSN 2351-7204 PETRAS SKIRMANTAS. Kol nebuvo Palemono: kompiuteriai, lietuviški rašmenys ir šriftai | 13 doi.org/10.35321/bkalba.2019.92.09 family, three more families of exactly the same kind were added under expanded names: LT-Times-GR, LT-Times-AK, and LT-Times-CF. Here, GR indicates the grave accent, AK the acute accent, and CF the circumflex. In the correspondingly named fonts, the relevant accent was added—drawn above—all Lithuanian letters that can bear it. This produced a family of 16 fonts, allowing the stress of letters to be changed just as their weight or italicization can be changed. This made it easy to add stress marks to text that had already been set, then convert stressed text to unstressed at a single stroke, or very simply change one type of stress mark on a letter to another, and so on. Such stress marking does not interfere at all with spell-checking or automatic hyphenation; the auxiliary programs designed for those tasks do not even see it. Nor would they see it—for stressed and unstressed letters in fact differ only in their shapes, the visual forms displayed on a screen or printed, while inside the computer they are the same and are represented by the same codes. To put it more clearly, stress marking here is not a property of the text itself, but of its display in certain programs, such as MS Word or page-layout programs, much like character formatting. All this was suitable and quite convenient for preparing and publishing linguistic works. For example, stress marking required no changes to the computer keyboard, since text was marked for stress using the character-formatting tools available in MS Word, supplemented by custom tools based on VBA applications (scripts). Among other things, this method of marking stress somewhat recalled the approach taken in linguistics to stress and pitch accent: that they are properties or features of syllables, not sounds, are not to be confused with sounds, and belong to a separate, suprasegmental level. With this method, the marks indicating stress and pitch accent are not assigned to the text but raised to another level—in effect, the formatting level—where they are merely placed above the letter marking the center of the stressed syllable. The letters in both stressed and unstressed text are the same “inside” the computer and are encoded identically. The almost universally used and frequently repeated phrase kirčiuotos raidės (“stressed letters”) is not very precise, since letters are graphic counterparts of sounds; therefore, consistently following the aforementioned view, it would often be more appropriate to say kirčiuoti skiemenys (“stressed syllables”). Unfortunately, in computer writing and print there are no tools that would allow stress marking to be associated specifically with syllables rather than with individual letters. The method of representing stress marks described above, which was suitable for the “paper” publishing of the time, also had a major drawback: by moving stress marking to the formatting level, it was in principle impossible to represent stress wherever there was no formatting! In other words, there was no way at all to represent stress in so-called plain (English plane) text. Therefore, this method of stress marking was unsuitable for anything other than preparing printed publications. Its practical scope was indeed narrow and limited. And of the set of 16 fonts in the LT-Times collection, only four of one BENDRINĖ KALBA 92 (2019) www.bendrinekalba.lt ISSN 2351-7204 PETRAS SKIRMANTAS. Kol nebuvo Palemono: kompiuteriai, lietuviški rašmenys ir šriftai | 14 doi.org/10.35321/bkalba.2019.92.09 family fonts—those in which all Lithuanian letters were unaccented. The other 12 fonts were simply auxiliary modifications of those four fonts, intended to convey the described method of rendering accents. Incidentally, the latter disregarded one of the greatest taboos in this field—the inviolability of the ASCII portion: accent marks were inserted above the single a, e, i, o, u, l, m, n, r as well, which belong to the standard character set and are encoded with ASCII codes... Other sets of special linguistic characters needed for publishing were compiled in a similar way—or, more precisely, fonts intended for those sets: the basic characters of a set were placed in one, initial font, while their variants, supplemented with auxiliary marks, diacritics, and so on, were placed in the other fonts in that set. Each such special character was generally assigned its own font. Unsystematic, random, very rare, or simply unusual special characters that did not fit into any set were placed in a separate auxiliary font, often called Varia. In time, more than one such Varia font began to appear; names had to be supplemented and changed, and before long it became customary to create a similar auxiliary font for practically every substantial linguistic work or publication. Almost at the same time as accented letters, computers were also found wanting in the phonetic transcription characters used by Lithuanian linguists. They were needed, of course, for both Kalbotyrai and Baltistikai, but the need was nevertheless unsystematic: they were required only from time to time, and auxiliary fonts could be used, with a separate Varia created for each volume. After a brief pause following 1990, publication of those journals resumed in 1993–1994 and began to gather pace again. By late 1994 and in 1995, they no longer contained any hand-drawn accent marks or other handwritten characters. At about the same time, the late Professor Aleksas Girdenis, the inspiration, supporter, and true spiritus movens behind many of the developments described here, was preparing for publication a book of texts transcribed in his native dialect. As the work neared completion, the question arose: how could it be published, laid out, and printed? The work used a great many dialect transcription characters of a kind found on no computer, and most of the book’s text was written entirely in them. You could not even offer such a text to publishers! It became necessary to create a special set of fonts that could display the required dialect transcription characters. Vector drawings of the characters were prepared from the book’s author’s sketches and oral descriptions; codes were assigned to them at the creators’ discretion and “without taboos”—from both the ANSI and ASCII zones. Thus a font containing the basic phonetic transcription characters came into being. Copies of it were modified, with additional marks and diacritics attached to the basic characters, each conveying a different shade of pronunciation of the sound represented by the character. The result was a set of eight computer fonts containing phonetic transcription characters, each font conveying different shades. Since the professor’s native dialect is BENDRINĖ KALBA 92 (2019) www.bendrinekalba.lt ISSN 2351-7204 PETRAS SKIRMANTAS. Kol nebuvo Palemono: kompiuteriai, lietuviški rašmenys ir šriftai | 15 doi.org/10.35321/bkalba.2019.92.09 Samogitian, and, naturally, the book included transcribed speech examples he had recorded in his native area, the character set was named Zemtra, taking the beginning of the words Žemaitiška transkripcija (“Samogitian transcription”). Using it together with the previously discussed font set LT-Times, the book of transcribed Samogitian texts was typeset, laid out, and submitted to the publisher, which brought it out in 1996 under the title Taip šneka tirkšliškiai (“This Is How the People of Tirkšliai Speak”). All this showed that the possibilities for preparing books for print using a computer were truly extensive and flexible. Producing such a book using “metal” printing technology would have posed immense problems and might not have been possible at all: most likely, the set of transcription characters would have had to be narrowed, adapting the transcription of speech to the printing house’s capabilities. Now, however, the characters and their variety in the book could be tailored to the author’s ideas and choices. In other words, characters could now be adapted to requirements, providing whatever characters were needed. Previously, requirements had to be adapted to the characters available. It may be assumed that this carefully prepared book of transcribed texts by A. Girdenis also served as an impetus and model for creating similar books of texts from other dialects, subdialects, and local varieties. The number of linguists taking up this work grew, and the Institute of the Lithuanian Language became actively involved. The geographic range of examples of living speech recorded in phonetic transcription expanded rapidly, encompassing various regions of Lithuania, and the characters in the Zemtra set, adapted for transcribing the speech of Northern Samogitians, began to prove insufficient; at least, not all of them were suitable in every case. It was possible, though inconvenient, to manage by combining Zemtra’s characters with those from other home-made fonts. The Institute of the Lithuanian Language created its own fonts containing characters for transcribing dialects and for other purposes (TimesDialect0 – TimesDialect2, Alt, Altt etc.), while new supplementary Varia... fonts were continually being created at the university. In short, it became clear that a different set of fonts for phonetic transcription was needed—one whose characters could express the phonetic subtleties of any Lithuanian dialect or local variety. Professor A. Girdenis selected J. Gerulis’s transcription as the basis for a new set of dialect-writing characters, supplemented it, and clearly specified how each phonetic character should look. Vector drawings of those characters, assigned codes once again at the creators’ discretion and “without taboos,” were placed in ten fonts, identified by numbers from 0 to 9, and the entire set was named Fontra, combining the beginnings of Fonetinė transkripcija (“phonetic transcription”). Fontra included essentially all the characters in Zemtra and had others needed to represent the speech of other regions. This set greatly facilitated the preparation and publication of dialect-text books and other linguistic publications, which were now gathering pace. In 1998, an article explaining how Fontra was compiled, what it was intended for, the purpose of its characters, and their usage characteristics was published in Kalbotyroje (vol. 47, issue 1, pp. 107–121). Having displaced Zemtra, BENDRINĖ KALBA 92 (2019) www.bendrinekalba.lt ISSN 2351-7204 PETRAS SKIRMANTAS. Kol nebuvo Palemono: kompiuteriai, lietuviški rašmenys ir šriftai | 16 doi.org/10.35321/bkalba.2019.92.09 Fontra remained in use for quite a long time, almost a decade—until 2004–2005, when it began to be displaced by the modern, fundamentally different Palemonas. Toward the end of the previous millennium and the beginning of the new one, the number of books printed using phonetic transcription and transcription symbols grew considerably; they began to be published one after another, and Fontra probably had some influence on this. Incidentally, Professor A. Girdenis was very fond of Fontra, considered it an effective tool for phonetic transcription, best suited to printing transcribed dialect texts, and wanted the texts he prepared to be printed in it. A. Girdenis did not take to Palemonas... Therefore, the second book of dialect texts prepared by the professor, Žemaičių dzūkai, published in 2008, was printed in Fontra, although by then Palemonas had already replaced it in many places. Only the third collection of Žemaitian texts, Šiaurės žemaičių Skuodo zona, published after the professor's death (†2011), was converted by his students to Palemonas... It was printed in 2013. The third set of fonts conveying the special characters needed for linguistic work, assembled in the same way as the two already described and named IndoBalt, was intended for the distinctive letters and symbols found in works on Indo-European studies, Baltic studies, and historical grammar. It greatly facilitated the preparation of texts for Kalbotyra and Baltistikos and was used in publishing other linguistic works. With the three aforementioned font sets—LT-Times, Fontra, and IndoBalt—the Faculty of Philology at Vilnius University no longer felt a shortage of the special characters required for linguistic work, especially since an auxiliary font, another Varia..., could easily be made when needed. The Faculty of Philology then published almost none of the written heritage; this work was carried out largely by the Institute of the Lithuanian Language, and so the problems associated with heritage-specific characters fell on its shoulders. They were resolved in much the same way—by creating fonts with the required characters. This necessity-driven practice of “homemade” font creation, reminiscent of the extremely popular “do it yourself” attitude of the not-so-distant Soviet era, made it possible to cope fairly successfully with the challenges of preparing and publishing linguistic literature that arose from changes in printing technology brought about by the development of computers. It became possible to prepare and publish linguistic works of almost any typographic complexity. But drawbacks also began to emerge: homemade fonts proliferated, the overall number of special characters grew, and characters began to duplicate and recur... Keeping track of and managing the ever-expanding inventory of fonts and characters was becoming difficult. Text entry slowed down: there were simply not enough keyboard keys for all the characters, and one often had to rummage through an increasingly thick folder of character tables in search of some more unusual letter... It was particularly problematic to transfer and send texts that used homemade “special-purpose” fonts from one computer to another. After all, every font used in the transferred BENDRINĖ KALBA 92 (2019) www.bendrinekalba.lt ISSN 2351-7204 PETRAS SKIRMANTAS. Kol nebuvo Palemono: kompiuteriai, lietuviški rašmenys ir šriftai | 17 doi.org/10.35321/bkalba.2019.92.09 text had to be installed on the computer. In practice, this meant that the text being prepared or already completed—the computer document file—had to be sent together with the set of fonts used in it, which was not a good solution: who would want to install several, or sometimes even a dozen or more, uncertified fonts of dubious origin and quality on their computer just to read one or two documents correctly? Meanwhile, computer networks were expanding, the internet was becoming ever more functional and pervasive, and electronic dictionaries began to appear, distributed both on media and online, followed somewhat later by electronic books. Alongside traditional publications issued in print, electronic versions began to emerge. Linguistic publications began moving in that direction too. And homemade specialized fonts with unrestricted, “no holds barred” encoding, perfectly suited to traditional publishing, were becoming a major and extremely difficult obstacle. Software designed for working with texts also underwent changes. From its eighth version, better known as Microsoft Word, Word 97, the most popular writing tool and the one most familiar to philologists allowed the use of computer fonts with far more than 255 characters: as Unicode, by then established and gaining authority, provided, they could contain hundreds or thousands. The fonts in the homemade sets described above were not like that, but the emerging prospect of large fonts seemed attractive. After all, a single such font could contain all the special characters needed for philologists’ work! There would still be room for the ordinary, traditional letters of various alphabets and their variants, since a font in this format, though not limitless either, could hold more than 65,500 different characters. This prospect became more realistic still with the new print-layout program Adobe released by the company at the turn of the millennium, InDesign, which also supported Unicode character encoding and fonts based on it (all versions of the previously used layout program Adobe PageMaker, except the last, seventh version—which was released after InDesign appeared and soon disappeared—recognized only ANSI-type codes). The idea of a single font containing all the characters needed for linguistics was further advanced by the fact that dialect texts transcribed on computers began not only to be published in books, but also to be stored in computer repositories and databases: the abundance and variety of fonts with phonetic transcription characters were turning into a jumble, while the freedom to encode characters was leading, quite frankly, to chaos. This, naturally, constrained the development of character sets and raised doubts about their functionality and usefulness. In short, the need for a new, multifunctional font—that is, one suitable for print, the internet, and databases—was felt everywhere. The State Commission of the Lithuanian Language took the initiative to bring such a font into being. In autumn 2002, the working group it had convened to create a new font gathered for its first working meeting. And in autumn 2006, during the Infobalt 2006 exhibition, completed and tested, and essentially already used in books since 2004 BENDRINĖ KALBA 92 (2019) www.bendrinekalba.lt ISSN 2351-7204 PETRAS SKIRMANTAS. Kol nebuvo Palemono: kompiuteriai, lietuviški rašmenys ir šriftai | 18 doi.org/10.35321/bkalba.2019.92.09 The new typeface Palemonas used for publishing was presented to the public. The group that created Palemonas also prepared and published in 2005 the book Lituanistinis šriftas Palemonas, most of which consists of specific technical information—lists of characters grouped by base letters, with each character’s Unicode name and encoding indicated, as well as diagrams that help locate the required character quickly. Palemonas is perhaps the only Lithuanian computer typeface documented in such detail. While work on Palemonas was under way, around 2003 a predecessor appeared—Fotonijos commercial typeface Aistika, created in the same format; it was publicly presented as the first Lithuanian computer typeface. Although this claim may be doubted (were the previously distributed Fotonijos TimesLT ir TimesLTAccented not Lithuanian?), Aistika was still the first truly successful work in this field, embodying Lithuanian characters—not only the usual ones, but also a considerable number of special characters—in a modern Unicode-encoded typeface. Aistika was intended for other Fotonijos products, the computer dictionaries it distributed, but it also enabled high-quality writing in both standard and accented Lithuanian or Russian letters, and included Greek characters, numerous transcription marks, and other symbols. However, Aistika, too, did not provide all the characters required for philological publications. Therefore, Palemonas remained necessary even after Aistika appeared, especially since it had been conceived as a free, freely distributable typeface, and remains so to this day. These two Unicode-encoded typefaces are compatible with each other, and the encoding of all accented letters (even those without Unicode codes) is also coordinated. Thus both ordinary and accented Lithuanian text set in either one can be rendered in the other without distortion. With the creation of Palemonas, the problem of special characters required for linguistic and Lithuanian studies was essentially solved. Received 2019 12 15 BENDRINĖ KALBA 92 (2019) www.bendrinekalba.lt ISSN 2351-7204 PETRAS SKIRMANTAS. Kol nebuvo Palemono: kompiuteriai, lietuviški rašmenys ir šriftai | 19 doi.org/10.35321/bkalba.2019.92.09 BEFORE THERE WAS PALEMONAS: COMPUTERS, LITHUANIAN GRAPHEMES, AND FONTS Summary The article is an account of the difficulties that had to be overcome in preparing philological publications on Lithuanian linguistics, containing special graphemes (beyond the standard Lithuanian alphabet) in the last decade of the 20th century, when the typesetting technologies underwent radical changes with the advent of computer typesetting. An emphasis is placed on the importance and possibilities opened by the graphical interface in the IBM PC-type computers and the vector fonts that came with it. The topics discussed include character encoding schemes current at the time and the situation of the Lithuanian alphabet therein. The main focus of the article however is on the special linguistic characters used in Lithuanian philology (beyond the standard Lithuanian alphabet letters) and the technical challenges associated with them. The topics discussed include in-house-made special computer fonts and collections of fonts, used at the University of Vilnius Faculty of Philology at the time, for the publication of the periodicals Baltistica and Kalbotyra as well as linguistic books containing phonetically transcribed texts in Lithuanian dialects, with Lithuanian accentuation (syllable tone) marks, the so-called ‘Gerulian’ phonetic alphabet (i. e. phonetic transcription system based on the so-called Copenhagen Scheme of Transliteration and Phonetic Transcription and introduced into the study of Lithuanian dialects by a prominent scholar Jurgis Gerulis) as well as other characters necessary for this type of publications that were not available in the standard computer fonts of the time. It must be admitted that the sphere of usage of these in-house-made fonts was rather narrow and confined to publishing of linguistic texts on paper; thus, these sets of fonts were not a universal instrument of information exchange and this circumstance rendered them inferior to other, standard fonts. Difficulties arising from this non-universality and the proliferation of varieties of these privately made fonts are also discussed; the issue especially came into prominence when the time had come to publish linguistic texts on the internet and to assemble them into sortable and searchable databases. These difficulties and the new demands arising thereof are the main reason for creating a modern, universal computer font that should encompass all graphemes necessary for linguistics texts, especially those of Lithuanian linguistics. BENDRINĖ KALBA 92 (2019) www.bendrinekalba.lt ISSN 2351-7204 PETRAS SKIRMANTAS. Kol nebuvo Palemono: kompiuteriai, lietuviški rašmenys ir šriftai | 20 doi.org/10.35321/bkalba.2019.92.09 KEYWORDS: font, special characters for linguistics, encoding, character set, code page, encoding standards, ASCII, ANSI, Unicode, accent notation in writing, phonetic transcription PETRAS SKIRMANTAS Institute of the Lithuanian Language 5 Petro Vileišio St., LT-10308 Vilnius [email protected] BENDRINĖ KALBA 92 (2019) www.bendrinekalba.lt ISSN 2351-7204