Development of standards for sinhala computing

1st Regional Conference on ICT and E-Paradigms24th – 26th June 2004, Colombo, Sri Lanka Development of Standards for Sinhala Computing
Gihan Dias and Aruni Goonetilleke Abstract
Sinhala speakers. A similar initiative, not covered in thispaper, is being carried out both in Sri Lanka and abroad, Information technology has been used in Sri Lanka for about 20 years, but to a great extent, its use has beenlimited to those with a knowledge of English. And as theseare a minority of the population, the larger proportion of 2. Local Language Support
Sri Lanka’s citizens has not been able to make use of theinformation revolution.
Requirements
A number of initiatives to introduce Sinhala computing A number of inter-related elements are required to support have been made from the 1980s. We outline the progress a given language in a computer system. These are: of this work, and describe the work done by CINTEC and Character Encoding: how letters and words are
represented in a system. Each letter or other symbol is The encoding of Sinhala in a standard manner to facilitate represented by a code. For documents to be portable across information interchange, the development of Sinhala fonts, systems, they must be encoded in a standard format. For and Sinhala keyboards are described. We present the example, the ASCII code is one (but not the only) method rationale of the design, and not simply the final standard.
Fonts: how text is represented on a screen or printer. The
same character may be written in several ways, with scope 1. Introduction
Currently most computer operating systems, databases and Text input: from a keyboard, pen, voice recognition
applications in Sri Lanka work only in English. However, system, etc. The most common text input method is the the majority of Lankans are more familiar with Sinhala or keyboard. Both the keyboard layout, i.e., the assignment of Tamil, and prefer to use IT in their own language. This has keys to letters, and key sequences, i.e., what sequence of resulted in a gap between how people want to use keys yields a given character, should be defined.
computers, and what today's computers can do.
Application support for the language. In addition to text
Although the corporate sector in this country operates handling, each application may have local language menus, mainly in English, small businesses and the government work mostly in Sinhala or Tamil. Individuals usecomputers for personal work, such as e-mail, at home or in Utilities: such as spelling checkers.
telecentres. Lankans working abroad need to communicatewith their relatives in Sri Lanka. The dispersion andeffective use of IT in these sectors require that they support 3. Review of Current Sinhala Technology
A number of Sinhala fonts and applications are currently Countries such as Japan, Korea and Thailand, among others, had similar problems when they started usingcomputers. However, Japanese, Korean or Thai language is fonts – which may be used with any application and
now standard on computers used in those countries, and it packages which bundle an application (e.g. a word
is common for people in those countries to use computers and applications in their own language.
Although these systems are disparate, many of them share Why then didn't local-language computing become common in Sri Lanka? One reason is that unlike in theabove countries, a number of Lankans, including many 8-bit character set: Almost all current fonts are based on
decision and opinion makers, are proficient in English.
an 8-bit character encoding, which limits them to less than Also, people assume that using IT requires a knowledge of English, and have not demanded a change. Another reasonis the small size of the Lankan market.
Character codes based on keyboard layout: Many
current fonts map Sinhala symbols to the codes used by This situation is now changing. A significant number of Roman letters in ASCII. The two most common mapping non-English-speaking persons want to use Information schemes are based on the Wijesekera keyboard layout and Technology (IT). This is not limited to the use of the “phonetic” layout (i.e., where Sinhala letters are placed computers, but devices such as phones, game consoles etc.
on the same keys as their English sound-alikes, e.g. ක is onthe 'k' key and This paper documents the initiatives, by a number of persons and organisations, to make IT readily available to Consequently, the codes allocated to Sinhala letters are based on the codes of the corresponding English letters. In 4. Review of the Development of
a “phonetic” mapping the letter ක has the code Hex 6B, Computing in Sinhala
which is the ASCII code for the letter k. In a Wijesekeramapping, on the other hand, ක is on the key for 'l' (el), and A number of efforts were made to introduce Sinhala is therefore assigned the code Hex 6C (ASCII for l).
language computing in Sri Lanka. Two pioneering efforts In addition to the vowels (අ, ඉ etc.) and consonants (ස, ජ by DMS and Metropolitan in the 1980s did not gain etc.), the kombuva ( )and other symbols ( , , etc.) are widespread use. A patent dispute between these two each allocated a key and a code. Symbols such as the al- companies threatened the development of Sinhala language lakuna, ispili and paapili are displayed above or below the computing, but was eventually settled out of court. It previous character. A Sinhala letter is represented by one should be noted that bit-mapped displays and printers were or more codes; e.g., ( ක = + ක, ක = + ක + , not in widespread use at that time, and both these systems As Sinhala letters are of differing widths (and heights), Thereafter, Wijeya Graphics produced a Sinhala font for some systems provide two or more pili of differing widths, Macintosh, which was widely used in publishing. The to be used with various letters, e.g., ප and න. Also, some University of Colombo developed a Sinhala screen output for television displays that was used to provide election ක, ර and ළ, take non-standard forms, and are thus represented by individual codes.
result displays in the three languages Sinhala, Tamil andEnglish. Use of an 8-bit character code, especially one using the same codes as English letters for Sinhala symbols, 4.1 The Sinhala Alphabet and Alphabetical
simplifies usage as applications need no modification touse Sinhala. As far as an application is concerned, it deals with the same codes in Sinhala as in English.
The requirement for a standard code was identified in the mid-eighties and the Computer and Information 3.1 Issues with existing Sinhala packages
Technology Council of Sri Lanka (CINTEC) together withthe Natural Resources, Energy and Science Authority of Although existing Sinhala packages serve their intended Sri Lanka (NARESA) formed the Committee on purpose, i.e., allowing Sinhala documents to be produced, Adaptation of National Languages in IT (CANLIT), which and to use applications in Sinhala, they face a number of agreed on a unique Sinhala alphabet and alphabetical order. No immediate action was taken on Tamil, due to the Lack of needed letters: Due to the constraints of the 8-bit
encoding, some existing systems do not support all Sinhalaletters, e.g. CANLIT defined the Sinhala alphabet as having 16 ඍ (iru-yanna). In addition, due to keyboard limitations, some letters need to be entered using function vowels, 2 semi consonants and 41 consonants as shown in keys, keycodes, or the “insert symbol” feature.
the CINTEC publication of 1990 [5]. 13 consonantmodifiers were also identified. A new character to denote Problems with Pili: some systems provide only one keti-
“fa” (ෆ) was introduced. CANLIT also agreed on the ispilla, for example, which is used for both tall letters (e.g., alphabetical order with a slight modification. This exercise ව) and short ones (e.g., ග). Such fonts are not visually took a representative group of language and technology attractive. A similar problem arises with the length of the experts several months to arrive at a consensus solution pili. Another problem is the separation of letters andassociated pili when intra-word spacing is increased for A standard Sinhala encoding, known as SLASCII, was approved by the Sri Lanka Standards Institute as SLS 1134in1996 [3]. SLASCII was the basis for, but differs in many Collation and Searching do not work properly, as the
aspects with, the Unicode for Sinhala approved later in encoding of letters is not based on their collation sequence.
Also, the use of separate codes for the kombuva and pilicomplicates the collation algorithm.
4.2 Unicode Compatible Sinhala Code
3.2 Non-Standardisation
A standard encoding known as Unicode has been developed to handle all the world's scripts. Unicode The most significant problem with existing systems, includes provision for many languages, including Sinhala however, is not any of the above, but the lack of a and Tamil, among other Indic languages. Most modern standard. Text entered using one system must be read using computer systems, including Windows XP and Unixes, the same system. Even though most Wijesekera-based systems use similar encodings, they differ in certaincharacters. It has therefore become mandatory to send a The existence of a draft code for Sinhala proposed to the copy of the font together with a Sinhala document, unless ISO 10646 / Unicode Working Group, by researchers one knows that the recipient already has the font. This has based in Europe was first brought to our notice in the late made Sinhala e-mail impracticable, and slowed the use of eighties. This distorted the Sinhala character set with several glaring errors and omissions. For example, theletters ඇ and ඈ (which do not appear in other Indic The lack of a standard Sinhala keyboard not only causes languages), had been moved to the end of the character set.
difficulties for typists, but has made it impractical formanufacturers to provide Sinhala keyboards. This in turn In 1997, Sri Lanka submitted a proposal for the Sinhala has created a further roadblock for Sinhala computing.
character code at the Unicode working group meeting inCrete, Greece This proposal competed with proposals fromUK, Ireland and the USA. The Sri Lankan draft was finally accepted with slight modifications. This was ratified at the Sinhala implementations, which are all symbol-based, and 1998 meeting of the working group held at Seattle, USA decided that none of them met the above requirements. It and the Sinhala Code Chart was included in Unicode [4] was therefore decided to follow the Unicode standard, Version 3.0. SLS 1134 was also accordingly revised in which is identical to SLS 1134 and ISO 10646.
However, it was observed that the Unicode encoding suffered from a number of shortcomings, which 4.3 Sinhala Computer Keyboard
nevertheless can be rectified without modifying the As a large number of Sinhala typists were using the government approved Wijesekera Keyboard, CINTEC first • lack of encodings for bandi akuru such as ක, developed and obtained government approval for the“Extended Wijesekera Keyboard for Electronic • lack of encodings for the yansaya, rakaransaya and Typewriters”, the intention being the introduction of Daisywheel and Golf-ball electronic typewriters then usedas an interface for microcomputer output. This included the • lack of guidance on the use of multiple vowel new character fa (ෆ) and 3 other additional key positions.
This layout was again modified for use with a standard • lack of guidance on the encoding of non-standard 4.4 Recent initiatives
It was decided to document these as a revision of the SLS 1134 standard [2]. In addition, the sections of the standard In December 2002, CINTEC formed a committee on on definitions and the description of the Sinhala language The committee decided that work should be carried out in When doing so, the committee decided to avoid any four major areas in order to enable computing in Sinhala.
changes to the encodings specified in Unicode, and also to follow Unicode conventions whenever feasible.
• A standard encoding of Sinhala characters 5.1 Conjunct letters (බ ඳ අකර)
Unicode does not encode any conjunct letters, such as න.
Such letters are a shorthand for writing a pure consonant(i.e., a consonant with the intrinsic vowel removed) • Availability of standards-based applications and followed by another consonant. E.g. නද = න. Both representations may be used interchangably. It was The functions of this committee were incorporated into the therefore decided to use the Unicode zero width joiner ICTA Language Requirements working Group in January (ZWJ) code to indicate when a conjunct letter is to be used.
2004. The standardisation work has been carried out by the E.g., න + + ද = නද, න + + zwj + ද = න. The zero-width SLSI Sinhala Working Group since August 2003.
joiner is a special Unicode code which is used to indicatethat the two adjacent letters are joined.
The first three are described in the following sections.
Work on the fourth is still on-going.
The committee compiled a list of conjunct Sinhala letters 5. Encoding
5.2 Yansaya and Rakaransaya
An encoding is a standard for representing letters and A frequent critisism of Unicode is the lack of codes for the words as binary data. The committee agreed on the yansaya and rakaransaya. The Unicode documents fail to following requirements of the character encoding.
mention these two symbols. However, we realised that thisomission was deliberate, as neither of these symbols are • It must be able to represent all contemporary and Sinhala letters. Rather, they are abreviations for the letters ය and ර respectively, when they follow a pure consonant.
E.g., සත' is the conventional way of writing සතය and මත • It should facilitate collation and searching.
As such words are generally spelled using the yansaya or A language such as Sinhala may be encoded in several rakaransaya, it was initially proposed to represent a ය or a ways. For example, each symbol may be assigned a code; ර following a pure consonant using the relevant symbol.
e.g., ක may be represented as + ක + . Alternatively, However, some words, such as මලර ජ and පසය ල, do not each letter, such as ක , may be assigned a single code.
use the constructs. We were thus faced with twoalternatives: Unicode represents each vowel or consonant by a code, and uses another code for each vowel sign, which modifies • encode the common case without any special codes; a consonant; e.g., ක is represented by two codes, ක + e.g., ත + + ර = ත and use the code zero-width non- . This approach follows the linguistic structure of Indic joiner (zwnj) to indicate when the construct should not languages such as Sinhala, whose basic unit is a syllable be formed; e.g., ත+ + zwnj + ර = තර formed by following a consonant with a vowel.
• use the code zero-width joiner (zwj) to indicate when The committee considered the encodings used in existing the yansaya or rakaransaya should be formed;e.g., ත + + zwj + ර = ත, ත + + ර = තර.
The first alternative yields a shorter code sequence for the 6. Fonts and Font Display
more common case, and also follows the Unicodeconvention that the common case is encoded without A font is a representation of the symbols in a script, for special codes. However, the committee selected the second display on a screen or on a printer. Legacy fonts comprise a one-to-one mapping between codes and glyphs, i.e., theshapes displayed. However, the Unicode encoding for Keying the sequence ත + + ර would otherwise have Sinhala (and other Indic languages) use a sequence of automatically produced a ත, even if not desired by the codes to represent each glyph. Therefore, a Unicode user. As the recommended keyboard has specific keys for Sinhala font must be able to handle such mappings, the ' and 1, users would use these keys to generate the e.g., න + + zwj + ද + + zwj + ර = න.
yansaya and rakaransaya, and the above key sequence forproducing තර.
Newer font technologies can handle such many-to-one mappings. One such technology, which is supported by Using the zwj to produce the yansaya and rakaransaya, both the Windows XP and Linux platforms, is OpenType which are forms of conjunct letters, allows us to use the [7]. As this technology is widely used, we decided to same representation for all conjunct letters.
concentrate our font development efforts on OpenType.
5.3 Rephaya
One method of implementing a Sinhala font is to produce a glyph for each letter, such as ක , ක9 or The rephaya, though less common today, is used to number of such glyphs, if every possible letter is to be represent the letter ර preceding another letter. For example represented, will exceed 6000. The Lake House has the word කරම may also be written as කර . This construct implemented such a font, as it allows them to have full represents a pure consonant (ර) followed by a letter, and control over the appearance of each letter. However, such may be encoded similarly to a conjunct letter using a zero- width joiner; e.g., ක + ර + + zwj + ම = කර .
Another approach is to have a glyph for each base letter, The sequence රය presents an interesting case. It may be and to position the pili around it. This yields a very small written as රය, ර or යර [6]; e.g., ආරය, ආර or ආයර. The font. However, in letters such as ට or ද, the pili are not second form is represented by ර + + zwj + ය. The simply placed above or below the letter, but form an sequence for the third form was not obvious, but was integral part of the letter. Other letters such as ර and ර, finally defined as ර + + zwj + ය + + zwj + ය.
have the pili in non-standard positions. Additionally, sinceletters are of differing widths, the widths of the associated 5.4 A Model of Sinhala text
pili should also differ. For these reasons, although pilipositioning may be suitable for applications such as mobile When considering the above cases, it was seen that devices, it does not provide the quality required for print conjunct letters, yansaya, rakaransaya and rephaya all represent combinations of two or three letters, where the Therefore, most general-purpose Sinhala fonts take a initial letter or letters are pure consonants. Such hybrid approach, and include some glyphs with in-built combinations may also be written as separate letters. We pili, and some, e.g., the kombuva, constructed the following model for the encoding of all separate glyphs. This aproach leads to another problem, namely, that the kombuva visually precedes the consonant, • A base Sinhala letter; i.e., a vowel or a consonant, is but is logically stored following it, and that some vowel signs, e.g., in ක , surround the consonant on two sides.
OpenType cannot handle such cases, which need operating • A pure consonant, or a consonant with a vowel, is represented by two codes, one for the consonant, andone for the associated sign (vowel sign or al-lakuna).
6.1 Operating System Support
• All conjuncts are represented by a sequence of letters Older operating systems such as Windows 95 only joined by zero-width-joiner codes. All letters other supported 8-bit fonts. Newer systems, such as Windows 2000, support Unicode, but do not handle “complexscripts” such as Sinhala properly. Although Windows XP 5.5 Non-Standard Letters
does support complex scripts, it does not correctly handleSinhala.
Some Sinhala letters are written in non-standard forms, and are represented by separate codes in current fonts, e.g., Unicode support on Windows is provided by a module ළ. However, ළ is not a different letter, but simply the form called the Uniscribe Unicode Script Processor. We of the letter murthaja-la with a paapilla. Therefore, it is collaborated with Microsoft to ensure that this incoprorated the features needed for Sinhala. The versionof Uniscribe released with Office 2003 supports Sinhala, Some pili take special forms when combining with some and correctly positions the kombuva before its base letter.
characters. For example: ක + = ක, ර + = ර. In thesecases too, we took the position that the encoding should be Linux systems employ a number of text handling based on the logical vowel sign, and not the shape mechanisms. The Pango library provides Unicode support in Linux, and is expected to support Sinhala.
Microsoft produced an OpenType Sinhala font named “Iskoola Pota”. We worked with Microsoft to ensure thatthis font incorporated the features specified in the SLS1134 standard. This font is currently in Beta release.
7. Keyboard
keyboard model, and allow efficient typing.
Although text may be input to a computer in many ways, 7.2 Design Principles
such as OCR, handwriting and speech, the most commonmethod of text entry remains the keyboard. The 101-key After considering several options, the committee decided computer keyboard has, with minor variations, become the on the following principles in the design of the Wijesekera- standard keyboard which is used for many languages. Thus our objective in the keyboard area was to specify standardsfor Sinhala text entry on such keyboards.
1. Retain all common letters in the same places as the We identified four types of keyboards for further 2. The 1st-row (number-row) keys (except the left-most) to be mapped to the same numbers and symbols as in • The Wijesekera keyboard, which is used with both 3. Have only one form of the al-lakuna and each pilla on • “Phonetic” keyboards, in which key assignment is the keyboard (the typewriter keyboard has separate keys for the different forms of the al-lakuna thepaapili).
• Transliteration schemes, in which text is typed as a 4. Do not have any “half letters” on the keyboard (as the first part of a bandi akura). Bandi akuru to be • Consonant-vowel sequence keyboards, in which the constucted pressing a join key between the two consonant is typed first, followed by the vowel 5. The sequence of keys to be typed to produce a India has defined a keyboard layout, called Inscript, which character be the same sequence as in writing; provides a standard way to type all Indian scripts.
e.g., + ක + + = ක9; ක + 1 + = = ක.
Microsoft had developed a modification of Inscript forSinhala as well. This was a consonant-vowel sequence Although this requires more keystrokes than in a consonant-modifier method (e.g., ක + ඕ-modifier = ක9), It was adopted to retain compatibility with The committee considered this keyboard, and was of the opinion that it did not meet our needs. This position wasvalidated at the Sinhala Language in IT seminar held in The keyboard layout was designed based on the above [2].
June 2003, where it was unanimously decided to use the Most letters were retained on the same key as on the Wijesekera keyboard. Therefore, the committee decided to typewriter keyboard, but the symbols which were on the 1st initially standardise the Wijesekera keyboard.
row were moved to other keys. Although we would have Nevertheless, it was recognised that the Wijesekera liked to have used only the unshifted and shifted keys, as keyboard had several shortcomings, and that we should there are more letters and pili than the available design a more optimal Sinhala keyboard. Such a keyboard (unshifted+shifted) keys, we needed to place some symbols may well be a consonant-vowel sequence type.
It was also realised that the sequence of keys used to Possible alternatives were to place these symbols on produce a given letter was independent of the keyboard control- keys or alt- keys. The standard followed by most layout, and that several layouts (e.g., a Wijesekera-based European etc. keyboards, is to use the right-hand alt- (also one and a phonetic one) may use the same key sequence.
known as ctrl-alt or alt-gr) key to enter additionalcharacters, leaving the left-hand alt- key for applications[8]. We decided to follow the same convention.
7.1 The Wijesekera Keyboard
Infrequently used letters such as ඏ (ilu) and kundaliya, This form of keyboard was originally designed for manual typewriters. It has the feature that all of the dead keys, i.e.,where the carriage does not move when the key is struck, In the typewriter keyboard, sagngnaka letters, such as ඟ are one the extreme left of the keyboard, due to design and ඳ, were produced by pressing a special key before the requirements of manual typewriters. Therefore, the al- corresponding “ordinary” letter. In order to make it easier lakuna, ispili, paapili, etc. are all on the leftmost keys.
to remember, it was decided to assign each sagngnaka Also, as manual typewriters only support one language, no letter to the alt-gr state of the same key as its “ordinary” consideration was given for compatibilty with English letter; e.g., alt-gr-ග = ඟ. The total number of keystrokes to keyboard. For example, the number keys were shited to the type a sagngnaka letter is thus the same as in the typewriter right by two places, and the puncuation marks appeared in Keys were also assigned to the yansaya, ' rakaransaya, 1 We had two conflicting objectives in standardising the rephaya, ර and punctuation marks. We attempted to keep punctuation either in the same position as on either the US-ASCII keyboard, or the Sinhala typewriter keyboard, but 1. Retain compatibility with the Wijesekera typewriter were not completely successful. In particular, the single and double quotes are placed on the “Z” key, which is not 2. Retain compatibility with the English (US-ASCII) In addition, we needed to be compliant with the Windows 7.3 Other Keyboards
9. Conclusion
When designing the Wijesekera-based keyboard, it was Our initial objective in this subject was to increase the use accepted that better keyboard layouts and key sequences of Sinhala on the Internet, and thereby make the Internet exist, just as the DVORAK keyboard layout is better than more useful to non-English speakers. However, whilst the QWERTY layout. However, it was recognised that undertaking this work, we realised that our task was attemting to design a new keyboard layout at this stage broader. Our present objective is to make using computers, would not only consume more time, but also hinder and other devices such as mobile phones, in Sinhala as acceptance among current typists who are familiar with the convenient and obvious as it is in English.
Wijesekera keyboard. Therefore, the design of an“optimal” keyboard for Sinhala was left for a later stage.
We believe that our work in the standardisation of the Such a keyboard may well use a consonant-modifier character encoding and the keyboard will assist in this task, keying sequence, as used in the Tamil99 keyboard.
Some current fonts, e.g. Kaputa, use a “phonetic” The next step is to disseminate the language support, keyboard layout, in which each Sinhala symbol is mapped keyboards, etc. througout the country, and to build to the key of a similar-sounding English letter. This awareness among users, potential users, web developers approach has the drawback that some Sinhala letters, such and application developers, that this technology exists, and as ත, ද and ෂ, have no English equivalents. Also, the is advantageous. We have embarked on this process, and assignment of the pili is somewhat arbitrary.
have started training and awareness programmes, as well asapproaching hardware and software vendors.
Another approach is to use transliteration, in which a string of English characters (with or without capitals) is We have also initiated a parallel programme to identify, mapped to each Sinhala letter, e.g., chi = ච, bhoo = භ.
and address, issues of using IT in Tamil.
Samanala is one such scheme which has been used for The authors wish to thank the members of the CINTEC, many years. Many people use such schemes informally to Sinhala Fonts Committee, the ICTA Language send Sinhala e-mail, SMS, etc. to friends using Roman Requirements Working Group, and the SLSI Sinhala Working Group, as well as all others who provided We recognised that many casual Sinhala users, who do not use a pre-printed Sinhala keyboard, need either a phoneticor transliteration-based keyboard. We did not, however,attempt to standardise is such a keyboard at this time.
References
[1] V.K. Samaranayake, S.T. Nandasara, J.B. Disanayaka, 8. Application Support
A.R. Weerasinghe, H. Wijayawardhana, AnIntroduction to UNICODE for Sinhala Characters, To use Sinhala effectively, applications should support the UCSC Technical Report 03/01, University of language. At the base level, applications should correctly input, store and display Sinhala text, possibly in [2] Sri Lanka Standards Institute, Draft Sri Lanka conjunction with other languages. Conformance with Standard Sinhala Character Code for Information Unicode simplifies this task, as any Unicode-compliant application will handle Sinhala without modification.
http://www.fonts.lk/doc/sls1134.pdf , 2004.
We tested a number of common applications such as web [3] Sri Lanka Standards Institute, Sri Lanka Standard SLS browsers, e-mail, etc. with the current system, and found 1134:1996 – Sinhala Character Code for Information that they handle Sinhala properly. However other applications, such as Microsoft Word, exhibited strangebehaviour. Also, we observed that although Microsoft [4] The Unicode Consortium, The Unicode Standard 4.0, Access and Excel can handle Sinhala text, they had problems with sorting, etc. This is attributed to the http://www.unicode.org/standard/standard.html .
incompleteness of the current Sinhala support, and isexpected to be rectified soon.
[5] S.T. Nandasara, J.B. Dissanayake, V.K.
Samaranayake, E.K. Seneviratne and T.
The next stage is the localisation of applications, so that Koannantakool Draft Standard for the Use of Sinhala they display menus, help, etc. in Sinhala. A large amount in Computer Technology, CINTEC, March 1990.
of work needs to be done in this area.
[6] J. B. Disanayaka, අකර හ පල (Letters and Strokes) The third level is Sinhala-aware applications and utilities, which “understand” Sinhala, such as spelling checkers,optical character recognition (OCR), etc. The development [7] Adobe Corp., An Introduction to OpenType, available of such applications is expected to be accelerated by the at http://www.adobe.com/type/opentype/main.html .
[8] Michael S. Kaplan and Cathy Wissink, “Unicode and Keyboards on Windows”, 23rd Inter-nationalizationand Unicode Conference, March 2003.

Source: http://www.siyabas.lk/docs/sinhala%20standards.pdf