ISO/TC37/SC2/WG1 N77
Date of presentation 2001-08-13
Proposer BSI
Draft technical report:
Development and Application of ISO 639
in the identification, classification and
alphanumeric coding of the
world's languages
Contents
1 Introductory Note
2 Clarification of Terms and Categories
3 The Global Context
4 The Proposal
5 Towards a Global Public Resource
Bibliography
Appendix
1 ISO 639
Codes correlated with the Linguasphere Referential Framework (extract
A-H)
Appendix 2 Linguasphere Referential Framework
of 10 Sectors and 100 Zones
Appendix 3 Chart of the World's
Arterial Languages (printed as 2 pages in landscape view)
Appendix 4 Global Language Index
(sample extract)
1 Introductory Note
There is an established need for a standardised system
of codes for the tagging and identification of the world's languages. Variation still exists, however, in the form
of language codes used by different organisations and in different countries. The ISO 639 codes provide the base for
standardisation in this field, although they at present cover only a small
proportion of the world's languages.
These ISO language codes also exist in 3 different versions, the ISO
639-1 two-letter code, and the ISO 639-2/T and 639-2/B three-letter codes (as
designed for terminological and bibliographical use, respectively).
A fully classified inventory of the world's languages
and speech communities was published in 1999/2000, including a coded index of
over 71,000 names (Linguasphere Register
of the World's Languages and Speech Communities, see Bibiography).
The following proposal outlines how the 3 versions of
the ISO 369 codes may be unified as a single standard, and how the formal
linking of this standard with the Linguasphere zones of reference would create
an alphanumeric Global Identification
Code (GIC) with increased informational content and inbuilt protection from
error.
2 Clarification of terms and categories
2.1
Classification codes, identification codes and referential codes
A clear
distinction needs to be maintained among 3 forms of language code:
2.1.1 modifiable classification codes or
"relationship scale", recording proximities of interrelationship
among languages but subject to modification as research progresses;
2.1.2 fixed identification codes or "language tags", enabling individual languages to
be identified without ambiguity; and
2.1.3 a stable "referential framework", providing a meeting-point for the
correlation of classification codes and identification codes (as in the
proposed Global Identification Code).
2.2 Language names and umbrella names
A clear
distinction needs to be maintained between:
2.2.1 language and dialect names as applied to individual spoken and/or written
varities of language; and
2.2.2 umbrella names, often artificially created,
covering groups or families of related languages (the treatment of which has
not been presented in the following pages, for reasons of time and space).
2.3 Languages and dialects
The
continuum which frequently exists among adjacent forms of speech means that it
has always been difficult define the boundary between usage of the terms
"language" and "dialect".
The situation is eased by recognising that many languages are better
analysed and distinguished in terms of three (rather than two) layers of immediate
relationship. These layers are best
explained by reference to specific examples:
2.3.1 outer language, as applied, for example, to the totality of the Welsh language in all
its spoken and written forms;
2.3.2 inner language, as applied to the 3 major components of the modern Welsh (outer)
language: literary Welsh (as written, and progressively standardised, in recent
centuries); northern spoken Welsh (in north Wales); and southern spoken Welsh
(in south Wales);
2.3.3 dialect,
as applied to distinct varieties of written Welsh (e.g. Bible or
"pulpit" Welsh) or to local varieties of northern or southern spoken
Welsh (e.g. Anglesey Welsh in the north, or Pembrokeshire Welsh in the south).
2.4 Spoken languages and standard written
languages
It is
of great importance that a clear distinction be maintained between:
2.4.1 spoken languages and their dialects, which may also be written (in dialect
literature, or in phonetic transcriptions, for example).
2.4.2 standard(ised) written languages, which
have acquired a status independent of the spoken word but which may themselves
be spoken (in speech which is modelled on the written tradition of a
language). Part of the present proposal
is that the 2-letter codes of ISO 639-1 should be formally recognised as
designating the relevant standard written languages (e.g. en for Standard English), in contrast to the general coverage of
the 3-letter codes of ISO 639-2 (e.g. eng
for English in any or all its forms).
3 The Global Context
The objective of clearly identifying all the languages
and speech communities of humankind, regardless of their demographic size, is
today clear, attainable and of global importance.
3.1 The Twenty-first century perspective
At the onset of the twenty-first century, humankind is
aware of itself as a single planetary community, with means of instant global
communication and of increasing global planning and coordination. Languages are the key to that communication
and coordination.
For the first time, the languages of the world may be
viewed as integral parts of humankind's greatest and most fundamental creation,
the continuous global web or "linguasphere" of human speech and
writing.
Languages no longer need to be listed and catalogued as
a vast array of independent objects, belonging to rival and often warring
communities. They can now be viewed and
classified as integral parts of a collective human heritage.
The classification of languages has until now been the
preserve of erudite specialists, often tracking down words of ancient languages
in the pursuit of evidence about the human past.
Today, however, the classification of modern languages
has a direct relevance to the way humankind perceives and organises itself as a
single global and multilingual community.
Individual languages can now be perceived, not as the individual
creation and property of specific communities, but as mutable and interrelating
subsystems within a vast global kaleidoscope of words, grammatical rules,
speech sounds and elements of writing.
All languages have benefitted or may potentially
benfefit from the modern communications revolution in two fundamental ways:
·
The recording and global transmission of the spoken
word allow any spoken language to share the advantages previously reserved to
written languages, enabling even small speech communities to maintain worldwide
spoken contact.
·
The instant transmission and exchange of the written
word allow any written language to share the advantages previously reserved to
speech, encouraging even children to use writing (instant messages by phone and
computer, and e-mail) as an integral part of their social life.
3.2 The Need for the Identification of Languages
within a Referential Framework
Any system of linguistic classification needs to contain
an element of fluidity, in order to deal not only with the fundamental nature
of the linguasphere but also with a still expanding knowledge of its
complexity.
At the same time, it is necessary that the identifiable
written and spoken languages of human communities be clearly and unambiguously
catalogued and identified, from the international use of English or French to
the unique speech of an isolated village in central Africa.
It is important to be aware of this contrast between (a)
the need for fluidity in establishing and updating a sliding scale of
linguistic interrelationships, and (b) the need for stability in identifying
the individual spoken and recorded languages of humankind.
The primary objective in this field is therefore to
complete a standardised international system of identification codes for the
unambiguous tagging of all known forms of spoken and written languages, alive
or recorded from the past, and for the correlation of those fixed tags to a
separate scale of linguistic interrelationships.
4 The Proposal
4.1 The Institutional Background
The first comprehensive coded and classified inventory
of the languages and speech communities of humankind during the 20th
century was completed in December 1999 and published in Wales in 2000.
This Linguasphere
Register of the World's Languages and Speech Communities provides a
referential framework for the location and classification of over 22,000
identifiable varieties of speech and writing.
The Linguasphere Register is supported by a unique and expandable Index
of over 71,000 linguistic and ethnolinguistic names, each classified and coded
within the referential framework, using comprehensive scale of linguistic
interrelationships.
The agency responsible for compiling and maintaining the
Register is the Linguasphere Observatory (www.linguasphere.org), a
transnational research network devoted to the study and maintenance of
multilingualism. Conceived in Canada in
1983, the Observatory was established in France during the 1980's. During the 1990's, it has worked in close
collaboration with the University of London's School of Oriental and African
Studies, and has been directed from bilingual Wales since 1995, with scientific
support from Russia, India and the United States. See further details at the end of section 3 of this paper.
In July 2001, the BSI (British Standards Institution)
requested the Linguasphere Observatory to make a firm proposal for the
establishment of a standardised alphanumeric coding system covering all the
world's languages, based on existing and future codes of ISO 639 and correlated
with the referential framework and relationship scale of the Linguasphere
Register.
4.2 The Technical Background
The
ISO Alpha-2 and Alpha-3 Codes for the Representation of Names of Languages (ISO
639) are complementary in purpose and form to the Numeric-2 Code
employed for the Linguasphere Referential Framework (LRF) of the world's
languages.
ISO 639 provides 2-letter or 3-letter tags (or "standardised
abbreviations") for the identification of specific languages and groups of
languages, whereas the LRF provides 2-digit tags for a referential inventory of
the world's languages within 10 sectors (1st digit) and 100 zones (2nd digit).
4.2.1 ISO 639 Codes
The ISO 639
codes for a range of the most commonly encountered names of languages (and
groups of languages) are presented in the International Organisation for
Standardisation's Code for the
representation of names of languages (1998), an are available online at /http://lcweb.loc.gov/standards/iso639-2/langhome.html/.
ISO 639-2, originally devised for use in library systems, now exists in slightly divergent forms, known as ISO 639-2/T (terminology) and ISO 639-2/B (bibliographic). Although the 3-letter codes of ISO 639-2 could provide codes for 26x26x26 languages, limits specified in the standard currently restrict the creation of new codes to languages with a substantial body of literature. If rigorously applied, this restriction limits the more generalised use of the IS-639 codes, particularly in ICT usage.
As a result, some ICT users – including ministries and official agencies - have either made use of the SIL (Summer Institute of Linguistics) codes, or have developed their own coding systems, notably the OpenType specifications (OT) used in font and rendering technologies. Such variant codes have been developed in certain countries, including the UK, Sweden and Germany, which in some cases have caused clashes in bibliographic information interchange.
4.2.2 Linguasphere Codes
The
Linguasphere 2-digit code is defined for over 22,000 modern languages and
dialects, and for their historical forms where relevant, in the Linguasphere Register of the World's
Languages and Speech Communities (published in 2 volumes by Linguasphere
Press, Hebron, Wales 2000). The
Linguasphere code (digital Reference Framework, plus alpha Relationship Scale)
is discussed and exemplified on the Linguasphere Observatory website (http://www.linguasphere.org).
The fully
coded Linguasphere classification and annotation of the world's languages is
already available with limited access online (http://www.linguasphere.net)
and is to be made freely accessible as a public resource within the next year
(see section 5 below).
Appendix 1 to this paper displays the 2-digit tags of the Linguasphere Reference
Framework as an additional column in the listing of ISO 639, as exemplified by
the letters A-H in an alphabetical arrangement by English names of
languages. Appendix 2 lists and explains the Linguasphere numerical tags, with their linguistic
and/or geographical applications. Appendix 3 provides a table of the world's arterial languages (each reaching over
1% of the world's total population), with the relevant ISO 639 and Linguasphere
codes. Appendix 4 presents an extract from the proposed Global Language Index, which is
available as a starting point for the systematic extension of ISO 639 codes.
4.3 Formulation of the Proposal
It is proposed by the British
TS/1 Committee and the British Standards Institution that a standard Global Identification Code for all known
languages and speech communities be established, by the prefixing of the 2
digits of the Linguasphere Referential Framework (LRF) to the 2 or 3 letters of
ISO 639 codes and to the extension of those codes to cover all spoken and
recorded languages.
The purpose of
this proposal is NOT to create yet another method of coding, but to
enable existing ISO (TC/37) standards to work more efficiently and accurately,
and to be expanded systematically to cover all languages and speech communities. The following pages outline how the 3
versions of the ISO 369 codes may be unified within this single standard, and
how the formal linking of the ISO codes with the Linguasphere zones of
reference would create an alphanumeric Global
Identification Code (GIC) with compact informational content and inbuilt
protection from error.
4.4 Practical considerations of the present
proposal
Some of the
problems hitherto associated with the codes of ISO 639, and with language
identification in general, were discussed by Peter Constable and Gary Simons of
SIL International in their paper Language
Identification and IT: Addressing problems of linguistic diversity on a global
scale, presented to the 17th
International Unicode Conference (San Jose California, September 2000).
Their paper
proposes the extension of the present Alpha-3 system to cover all known
varieties of written and spoken languages in the world, following the example
established in the SIL's Ethnologue (14th
edition, 2000). This proposal allows
for the establishment of thousands of 3-letter codes to represent language
names, not necessarily related in form to those names, but fails to address
some of the fundamental problems of isolated 3-letter codes.
Constable and
Simons (page 15) recognise the Linguasphere
Register as "the only likely candidate" as an alternative to
their proposed SIL system.
4.4.1 Advantages of the proposed Global Identification Code
The prefixing
of the digits of the relevant LRF numeric code to a 2-letter or 3-letter form
of an ISO 639 tag for a specific language would create a combined alphanumeric
LRF/ISO tag or Global Identification Code.
This combined tag would assist in solving several existing problems of
identification and referential classification.
In comparison with the existing ISO or SIL tags, the combined LRF/ISO
tag would be:
·
more transparent, with the initial digit indicating
one of five major affinities or one of five continental areas (e.g. 5 =
Indo-European or 8 = native South American: see Appendix 2 to this paper);
·
more easily located, with the 2 digits indicating a
linguistic group or area (e.g. 53 = Slavic or 87 = Amazon: see Appendix 2 to
this paper);
·
more readily classifiable, together with the names of
other related and/or adjacent languages, either within the same sector (first
digit in common) or zone (both digits in common)
·
better protected against typographical error in the
citation of tags (each alpha component being tied to a specific numeric
component).
At the same time, the continued existence of a single series of
unambiguous ISO 2-letter and 3-letter codes to identify the languages of the
world would mean that the combined LRF/ISOtags could be abbreviated for
practical purposes by the optional omission of the LRF numerical prefix.
·
In such abbreviated usage, the invisible LRF code
would still underlie any IT usage of the 2-letter or 3-letter components.
·
The LRF numerical prefix would thus be available not
only to classify language codes as required but, very importantly, to serve as
a check against typographical error. A
mistyped 2-letter or 3-letter code would have only a 1% chance of matching the
correct numerical prefix.
Use of combined LRF/ISO tags would also open the way to a more
structured approach to the classification of "language names", which
may be used to indicate a wide variety of different categories of language name
with an identical form of 3 letters.
·
The existing ISO Alpha-3 codes may indicate either
the name of a specific standardised language or of a wider
"language" composed of two or more closely related spoken and/or
written languages (e.g. the Ashkharik and Arewmta varieties of Armenian, or the
Gheg and Tosk varieties of Albanian), or an historical and/or liturgical
language (e.g. Church Slavonic), or a grouping of languages of
undetermined dimension or nature (e.g. Athapascan languages, or
"other" Austronesian, or "other" Creoles and pidgins). See examples in Appendix 1.
In contrast, the use of combined LRF/ISO tags could be associated with
the reconsolidation and extension of the 3 existing lists of ISO tags, to
create a single, more coherent and explicit system.
·
In an increasingly internationalised world, it is
appropriate that alpha codes for specific languages should be based wherever
possible on the autoglossonym (or indigenous form of the language name) rather
than on the English name (where this is different).
·
In this respect, where ISO 639-2/B diverges from
639-2/T, the 2/T code is generally to be preferred (e.g. /eus/ in preference to
/baq/ for Basque, for which the autoglossonym is Euskara). It may be noted that the Language Register also gives precedence
to autoglossonyms.
Most importantly, it would be important to distinguish between the
application of LRF/ISO tags to specific
objects (i.e. standard written languages) as opposed to "fuzzy" phenomena (i.e.
non-standardised languages, or continua of closely related spoken languages or
dialects) or to referential boxes
created or identified for use in the classification of languages (i.e. language
groups, families or categories or areas of languages, including the
Linguasphere sectors and zones).
·
The most frequent use of tags to represent language
names in IT is for the identification of specific standard written
languages.
à
For practical purposes, a standard written language
may be defined as a language whose form is largely fixed by means of a system
of graphic conventions, established and exemplified by the publication of a
large corpus of texts (normally in thousands or more).
à
A standard written language may also have a spoken
form, modelled largely on the use of the written form, in the same way that a
spoken language may also have a written form, transcribing actual speech.
·
Most such languages are already provided for under
the ISO 639-1 Alpha-2 code, and it would be helpful if the use of combined
LRF/ISO tags of the form Numeric-2 plus Alpha-2 could be specifically confined
to the identification of standard written languages (including their standard
spoken forms, wherever these are modelled on the written language).
·
The potential number of Alpha-2 codes (26 x 26 = 676) is
adequate to retain the existing ISO 639/1 codes for standard written languages,
regardless of the prefixed digits.
·
If the linguasphere were one day to include more than 676 such
languages, then it would be possible to duplicate some alpha codes under
different digits.
·
A more stable solution, however, would be to limit the use of Alpha-2
codes to a closed list of all those written languages standardised before the
end of the 20th century.
·
This basic LRF/ISO tag "for the representation
of the names of standard languages" would be only one character longer
than the existing Alpha-3 tags, but would be considerably more systematic and
rich in information, and more secure against typographical error. Cf. 79zh for Standard Chinese (rather than
/zho/ or /chi/ for all forms of Chinese) or 55sq for Standard Albanian (rather
than /sqi/ or /alb/ for all forms of Albanian).
·
The initial 7 or 5 locates Chinese and Albanian
within Sino-Tibetan or Indo-European, respectively, and it would be useful to
produce a list of LRF/ISO tags classified numerically by the LRF digits
(alongside the existing ISO lists arranged by alpha code or by names of
languages in English or in French).
·
The alphanumeric form of LRF/ISO tags would be
readily identifiable as language codes within other text, as opposed to the
potential confusion of some existing Alpha-3 tags with real words (e.g. /bug/
for Buginese or /got/ for Gothic).
In contrast, LRF/ISO tags applied to "other
types" of language name (non-standardised languages, fuzzy continua or
referential boxes) could be based on the Alpha-3 codes of ISO 639-2/T.
·
This distinction between Alpha-2
and Alpha-3 codes would be useful in distinguishing standardised languages
within fuzzy continua of spoken languages and dialects, e.g. between the
varieties of standardised Norwegian (Bokmål = 52no or Norwegian Nynorsk = 52nn) and "wider" Norwegian
in all its forms (= 52nor, which has fuzzy boundaries within the continuum of
other spoken forms of Scandinavian languages).
There are a number of other more detailed points to be
considered in the design of any improved codes for language names (including
the treatment of historical languages, for example), which will would dealt
with within the fully developed presentation of the proposed LRF/ISO
system. The Linguasphere Observatory
looks forward to productive discussions on all aspects of the development of
ISO 639, with members of BSI and ISO, and beyond.
5 Towards a Global Public Resource
5.1 Progress towards a tripartite global
reference guide
The present proposal is designed to provide the key
element in the production of a fully coded and interactive global reference guide to
·
the languages and speech communities of the world,
·
their established linguistic relationships,
·
their global corpus of linguistic and ethnic names, and
·
their geographic positions and demography.
This global reference guide would take the form of a
freely available, independent and multilingual website, comprising 3
interdependent "panoramas".
These would be interdependent, with a common alphanumeric coding system
throughout (ISO-639 plus Linguasphere), and would be fully inter-referenced and
interaccessible at every point:
5.1.1 the Global
Index of the World's Languages and Speech Communities (or ISO 639/ Linguasphere
Index), presenting an alphabetical key to the identification and location
of all known written and recorded languages and dialects, and all varieties of
linguistic, ethnic and communal names.
This panorama, covering a total of over 71,000 names, is already
available in a first printed edition (but without ISO codes), as the Index to the Linguasphere Register. This progressively updated and expanded
edition will be opened to free public access and dialogue on the internet
within the next year. An extract from
this Index, covering names beginning G-, has been prepared and is now being
extended as part of the current proposal.
This will include existing and proposed additional ISO-639 codes.
5.1.2 the Global
Register of the World's Languages and Speech Communities, presenting a
comprehensive scale of linguistic relationships among the spoken and recorded
languages and dialects of the world, and their relevant speech
communities. This panorama is already
available in a first printed edition as the Linguasphere
Register of the World's Languages and Speech Communities, covering over
22,000 varieties of languages and dialects.
This edition will be opened to free public access and dialogue on the
internet within the next year, and will be progressively updated and expanded
online. Extensive extracts are already
freely available at /www.linguasphere.org/.
5.1.3 the Global
Mapbase of the World's Languages and Speech Communities, presenting a
cartographic survey of the location, distribution and interrelationships of the
world's languages and speech communities.
This panorama has already been developed by the Linguasphere Observatory
for Africa (linguistically the most complex continent in the world), in
collaboration with the London School of Oriental and African Studies
(SOAS). It has been printed as the
first sheet of the Linguasphere Mapbase
of the World's Languages and Speech Communities and is currently being
extended into southern Europe and western Asia, in collaboration with the
Languages of the World unit of the Russian Academy of Sciences (Akademia Nauk)
. The first African layer of this map
is viewable at /http://www.soas.ac.uk/Geography/LanguageMapping/home.html/. This same page on the SOAS website
illustrates how subsequent layers of the Linguasphere Mapbase will be
accessible by zooming, down to the layer of urban speech communities (as
already surveyed and published for over 300 minority languages of London, see
Bibliography below).
5.2 Applications of the tripartite reference
guide
This three-part electronic reference guide will serve as
·
a transnational reference system and educational
resource for teaching covering
à the
global complexity of humankind, as represented by the overlying diversity of
its languages and the divergent welfare and cultures of its individual speech
communities,
à the
underlying unity of humankind, as represented by a worldwide continuum of
multilingual communication and intercommunal identities (the
"linguasphere"), and
à the
establishment of comprehensive links with - and annotated signposts towards - a
vast range of other electronic sources on the languages, peoples and cultures
of the world;
·
a stimulus to innovative teaching and research,
including
à the
active investigation and surveying of linguistic and ethnic realities and
relationships, including the continuous updating and expansion of the global
reference guide itself;
à the
transnational observation and documentation, regardless of frontiers, of
- the actual and relative welfare
of all speech communities in the world,
- the movement and migration of
speech communities and their members,
- the formation and distribution
of minority urban speech communities,
- the incidence of all forms of
genocide and other forms of discrimination
among ethnolinguistic
communities;
à the
awakening of public interest in questions of the transnational and multilingual
heritage and origins of communities and individuals (the "languages of our
ancestors"). Linked to the growing
strength of public interest in genealogical research, this development may be of particular importance
in encouraging the development of bilingualism among first language
English-speaking communities (in danger of becoming the only communities
deprived of the advantages of bilingualism, in an otherwise multilingual
world).
5.3 The Linguasphere Observatory
The present proposals and products are the outcome of
many years research and development at the Linguaphere Observatory in Wales and
at its previous location in France as the Observatoire Linguistique. Created in 1983, after planning and
discussion in Quebec (at CIRB, the Centre International pour la Recherche en
Bilinguisme at the Université Laval), the Observatory was set up in Normandy as
a transnational research network devoted to the study and development of
multilingualism (under the honorary presidency of Léopold Sédar Senghor of
Senegal, and registered under the French law of association of 1901).
Among other linguistic activities, the Observatory was
responsible for two bilingual exhibitions on languages at the Centre Georges
Pompidou in Paris during the 1980's, with substantial support from the
Government of Canada. (These
exhibitions subsequently toured internationally, including London, Liège and
Lagos, and around the world to Canberra.)
Since 1995, the Observatory has been based in a bilingual area of west
Wales, under the directorship of David Dalby, where the Linguasphere Register of the World's Languages and Speech Communities
was first published at the turn of the millennium (1999/2000). Scientific support has been received from
Russia, France, India and the United States.
It is appropriate that the present proposals and
products should emanate from Wales, a country whose language has successfully
resisted and survived the successive invasion of its territory by two of the
most powerful languages in the history of the world, Latin and English. All speech communities now need to consider
their relationship to English, as a global lingua franca, and in this respect
the indigenous speech community of Wales has the longest experience in the
world, having faced the growing strength of its English neighbour for more than
one millennium. The cultural strength
and linguistic survival of the Welsh-speaking community offer an important
message of encouragement to small speech communities everywhere. English has a transnational role to play in
the world, along with other "arterial" languages, but should be
developed in the service of a multilingual global society, NOT as the medium of
a monolingual culture.
That the British Standards Institution in London (BSI)
and the University of London's School of Oriental and African Studies (SOAS)
should have given their support to the proposals and products of the
Linguasphere Observatory in Wales is also significant. At a time when countries around the world
are devoting resources to the study of a language associated with England, it
is appropriate that major public institutions in that country should devote
resources to the study and development of multilingualism and of the languages
of the world.
Linguasphere Observatory
and British Standards
Institution August 2001
Comments
on this paper, prepared at relatively short notice for the ISO TC/37 meeting in
Toronto, will be greatly welcomed, by post or by e-mail to
/research@linguasphere.net/.
A more detailed proposal will be prepared by the Linguasphere Observatory for
the beginning of 2002, including the orderly extension of identification codes
to all spoken and written languages, and the examination of procedures for
combining language codes with codes for countries and for scripts.
Bibliography
Baker, Philip &
Eversley, John, Multilingual Capital: the
languages of London's schoolchildren, Battlebridge Press: London, 2000
Constable, Peter and Simons, Gary (SIL), Language Identification and IT: Addressing problems of linguistic
diversity on a global scale, presented
to the 17th International Unicode Conference, San Jose (California),
September 2000.
Grimes, Barbara F.
(editor), Ethnologue: Languages of the
World (14th ed.), SIL: Dallas, 2000
ISO Code for the representation
of names of languages, ISO, 1998
Linguasphere Register of the World's Languages and Speech Communities (2 volumes), Linguasphere Press: Hebron (Wales), 2000
WORLD'S ARTERIAL LANGUAGES