Review of Various Projects against Open Echoes Criteria

Open Echoes has a methodology for selecting languages to support to ensure resources are target at those with the most and poorest first language speakers. This post is a quick review of progress of a small selection of Open Source and other projects against Open Echoes target languages.

Project	Debian [1] [1a]	SourceForge [2] [2a]	Mandrake [3] [3a]	KDE [4] [4a] [4b]	Google in Your Language [5]
Bengali	4%	0%	77%	21%	100%
Chinese, Mandarin	39%	13%	98%	82%	99%
Hindi	8%	1%	89%	61%	100%
Panjabi, Western	10%	0%	5%	38%	100%
Urdu	0%	0%	1%		100%
Amharic	2%		22%		100%
Telugu	1%	0%			100%
Marathi	2%	0%			100%
Tamil	9%	1%	48%	82%	100%
Hausa					51%
Somali	0%				96%
Burmese	0%				12%
Yoruba	0%				100%
Sindhi					73%
Gujurati	6%			0%	100%
Ibo
Oromo, Western Central					29%
Arabic, Sudanese Spoken
Javanese		0%			92%
Nepali	4%				93%

This demonstrates a lack of support for Open Echoes target languages, especially those with less than 100 million speakers, and confirms the basic premise of behind the project.

Notes:
[1] Debian 2005-11-19, Comparison with French Translation
[1a] Standard Arabic is 6%
[2] SourceForge Projects at 2005-05-12, Percentage comparison with French translation projects
[2a] Standard Arabic is 2%
[3] Mandrake Linuxtools 10.2 2005-02-12
[3a] Standard Arabic is 98%
[4] KDE GUI stats 2005-11-16
[4a] Standard Arabic is 44%
[4b] Indonesian is 9%
[5] Google in your Language 2005-12-04
[5a] Standard Arabic is 100%

Initial Consultation Feedback - 4 - One World Foundation Africa

One World Foundation Africa (http://www.oneworldfoundation.co.uk) is a UK registered charity working in London and in Africa to empower and promote the independence of the socially excluded. It aims to promote their independence through training, advancement of education, community support, advocacy, advice and research into their conditions to inform its intervention.

Feedback about Open Echoes confirmed that ICT is vital for NGOs working in the field and has changed their way of working. Being able to work in local languages would be useful.

There is also a high degree of volunteerism in Africa but members of the diaspora are more likely to seek to volunteer within their own industry than to offer help with a general initiative such as translation. Volunteers would also need assistance with technical terminology through provision of technical dictionaries and other resources.

Prioritising Languages for Development

Open Echoes works with the languages of the poorest people of the world. What methodology should be used to determing which languages are a priority when considering the objective of promoting development?

Firstly a measure of lack of development is needed. The measure used in this article is the UN's Human Development Index (HDI) from its 2002 assessment. This rates countries on a scale between zero, very underdeveloped, to one, very developed. The measure is based on multiple factors such as child education, morality and access to clean water. Like most economic statistics its assessment is based on country boundaries and assumptions about the role of the state in development. Language is no respector of boarders and so any language selection will be distorted by the use of the country based HDI. An HDI measure is not available for all countries, for example Afganistan, Somalia and Iraq, and in these cases a neighbouring country's HDI is used as an estimate, most likely too high an estimate.

Secondly we need some measure of the number of users of a language so that effort can be focused on providing support to the highest number of people. The measure used in this article is the number of first language speakers totalled across all countries. Figures are taken from the 15th Edition of Ethnologue [1]. We also only consider languages with over one million native speakers of which around 350 are catalogued. Speakers of course may not be literate but the assumption is made that speakers will become literate and that Open Echoes' work will play a small part in that process. Support for first language speakers is also a political issue as it may well conflict with state language policy in certain countries. Ignoring second language speakers will also cause many important trade languages, such as Swahili, to be excluded but again Open Echoes is an initiative to support more languages rather than less so this bias is a lessor issue.

Next we need a way of combining these two measures to produce a ranking from large poor speaking populations to small rich ones.

From experimentation with ranking countries by HDI and population, a low HDI needs to be given a much higher priority or weighting than a large population to get a sensible ranking. Without this countries with large and relatively or highly developed populations such Mexico and the USA appear too high in the ranking. Using (1-HDI) to the power of three (i.e. cubed) produces a sensibly weighted ranking, for example Burkino Faso with an HDI of 0.3 produces 0.7 x 0.7 x 0.7 = 0.343.

HDI is a country measure whereas the number of first language speakers of a language is not. Therefore a method of determining the HDI of a language is needed. Here the HDI of the 'main country' of the language is used, with 'main country' as defined by Ethnologue. So for the Spanish language the HDI of Spain is used rather than say the HDI of Spanish speaking Chile. This may seem odd as it tends to associate each language with its highest HDI or most developed speaking country but it results in the desired effect of ranking higher the most isolated languages and underdeveloped countries. For example if Spanish as a language is associated with the high HDI of Spain then it would be expected that Spanish would be a well supported language to the shared benefit of Spanish speaking Chile with its lower HDI. This applies even though the Spanish used in Chile differs from Spain.

So what are the results of the ranking exercise? Lets take two examples to illustrate the methodology - Bengali and Somali.

Bengali has 171.1 million speakers in all countries with 100 million in its main country Bangladesh. Bangladesh had an HDI of 0.51 in 2002 making it a medium rather than low ranking country in development terms (just 0.01 above the low category). So the Bengali language is rated as (1 - 0.51) cubed times 171.1 population = 20.53.

Somali has 12.7 million speakers and Somalia has a much lower HDI than Bangladesh of 0.36 (using Ethiopia's HDI as an estimate) making it a low ranking country. So the Somali language is rated at (1 - 0.36) cubed times 12.7 = 3.3.

So the Bengali language ranks above Somali even though Somalia is less developed than Bangladesh because of the much higher number of Bengali first language speakers.

So which are the highest ranking languages? Here are three top tens: the first top ten is drawn from all languages, the second ten is drawn from all languages with main countries in the least developed half of HDI rankings, the third ten is drawn from main countries with an HDI regarded as low.

All:
Rank, Language, Main Country, Popul., Rating
1) Bengali, Bangladesh, 171.1, 20.5
2) Mandarin, China, 873, 17.5
3) Hindi, India, 180.7, 10.8
4) Panjabi, Pakistan, 60.8, 7.9
5) Urdu, Pakistan, 60.5, 7.9
6) Amharic, Ethiopia, 17.4, 4.5
7) Telugu, India, 69.7, 4.2
8) Marathi, India, 68, 4.1
9) Tamil, India, 66, 4.0
10) Hausa, Nigeria, 26.2, 3.9

Lowest 50% HDI Only:
Rank, Language, Main Country, Speakers, Rating
1) Bengali, Bangladesh, 171.1, 20.5
2) Hindi, India, 180.7, 10.8
3) Panjabi, Pakistan, 60.8, 7.9
4) Urdu, Pakistan, 60.5, 7.9
5) Amharic, Ethiopia, 17.4, 4.5
6) Telugu, India, 69.7, 4.2
7) Marathi, India, 68, 4.1
8) Tamil, India, 66, 4.0
9) Hausa, Nigeria, 26.2, 3.9
10) Somali, Somalia, 12.7, 3.3

Low Category HDI Only:
Rank, Language, Main Country, Speakers, Rating
1) Panjabi, Pakistan, 60.8, 7.9
2) Urdu, Pakistan, 60.5, 7.9
3) Amharic, Ethiopia, 17.4, 4.5
4) Hausa, Nigeria, 26.2, 3.9
5) Somali, Somalia, 12.7, 3.3
6) Yoruba, Nigeria, 19.3, 2.9
7) Sindhi, Pakistan, 21.4, 2.8
8) Ibo, Nigeria, 18, 2.7
9) Oromo, Ethiopia, 8.9, 2.3
10) Nyanja, Malawi, 9.3, 2.1

Of course when it comes to providing support we should expect that Bengali, Marthi, Mandarin and others already receive significant support and so lessor used langauges such as Somali will benefit instead.

The next exercise is to identify the support to be provided and identify those languages with the least current support.

References:

[1] Gordon, Raymond. Ed. (2005).Ethnologue, Languages of the World (15th Edition). Dallas: SIL International

Object - Preservation and Revitalisation of Endangered Languages

One of Open Echoes' "Objects" is "assisting in the preservation and revitalisation of endangered Languages using ... [Open Source] Software" as part of a programme to give the world's poorest people freedom as the means to develop, a programme inspired by Amartya Sen.

In researching how to assess the feasibility of this preservation objective and to provide an outline approach for working towards it I have referred to David Crystal's book Language Death [1], particularly Chapter Five, 'What can be done [about language death]?'. It is taken as given that language death is a problem we should care about and a problem which contributes to lack of development in some communities in the poorest countries.

Language death is also an issue in highly developed countries such as the USA, Australia and members of the European Union not just in less developed ones. As Open Echoes works with the languages of the poorest peoples of the world many endangered languages will be beyond its remit. In these cases considerable common cause exists and much experience can be shared. It needs noting that many of the causes of language death and problems relating to revitalisation (practical, political and economic) are more extreme in the poorest countries.

Crystal provides six primary factors for revitalising a language. He states that an endangered language will progess if its speakers:

1) increase their prestige within the dominant community;
2) increase their wealth relative to the dominant community;
3) increase their legitimate power in the eyes of the dominant community;
4) have a strong presence in the educational system;
5) can write their language down;
6) can make use of electronic technology.

It is striking how these factors reflect Amartya Sen's basic premise that people need freedom to develop, the basic freedom to communicate and participate in civic society. It is also striking how making use of electronic techonology can support the other factors by providing a sense of cultural vitality and modernity. Crystal also quotes two other authors' lists which support the Open Echoes premis that technology is important in language vitality and promoting development.

Yamamoto's nine factors "that help maintain and promote the small languages" [2]:

1) the existence of a dominant culture in favour of linguistic diversity;
2) a strong sense of ethnic identity within the endangered community;
3) the promotion of educational programmes about the endangered language and culture;
4) the creation of bilingual/bicultural school programmes;
5) the training of native speakers as teachers;
6) the involvement of the speach community as a whole;
7) the creation of language materials that are easy to use;
8) the development of written literature, both traditional and new;
9) the creation and strengthening of the environments in which the language must be used.

Landweer's eight "indicators of ethnolinguistic vitality" for an endangered language [3] :

1) the extent to which it can resist influence by a dominant urban culture;
2) the number of domains in which it is used;
3) the frequency and type of code switching;
4) the existence of a critical mass of fluent speakers;
5) the distribution of speakers across social networks;
6) the internal and external recognition of the group as a unique community;
7) its relative prestige, compared to surrounding languages;
8) its access to a stable economic base.

Crystal also provides a list of steps a language revitalisation team should take. Many of the early steps such as engaging community members, conducting language surveys etc upto steps such as a process of standardisation of speech and writing and divising a publicly usable alphabet need to occur before any steps can be taken to provide localised software. It is interesting to consider how localisation of software into dialects and the creation of technical terminology could play a part in language standardisation.

Once a language has been broadly standardised technology can begin to be involved in the following of Crystal's revitalisation steps in the project:

1) reinforcing the use of the language in homes and other domestic settings;
2) expanding the use of the written and language in the public domain;
3) giving the language a presence in schools with the aim of making it a medium of instruction;
4) curriculum materials are written and published for child and adult use;
5) texts in the language, of general public interest (such as stories, poems, newpaper articles), are written and published;
6) principles are established to get the language recognised as an official regional language.

In conclusion their appears to be a consensus in this small sample of linguistic literature that language preservation and revitalisation, economic development and technology are interrelated. Crystal's revitalisation project steps also provide a clear framework for selecting the most appropriate software for localisation and a series of pre-requisite linguistic steps as a guide to selecting languages for localisation.

It is clear that languages with many speakers can also be in danger of losing domains of usage which in turn threatens their long term survival, the new technology domain being an important one. A limited number of languages may therefore meet the combined objectives of language preservation and of promoting development for a significant population size. However identifying such languages will be difficult as they are unlikely to appear on lists of endangered languages. Identifying those languages subject to hostile official policies such as assimilation could provide an alternative approach to identification, albeit a politically charged one.

The following is a list of relevant organizations taken directly from the appendix of Crystal's book.

Ad Hoc Committee on Endangered Languages
c/o Université de Québec à Montréal
CP 8888,
succ. Centre-ville
Montréal
Québec
H3C 3P8
Canada
email M366050 at er.uqam.ca

Committee on Endangered Languages and their Preservation (CLEP)
c/o Linguistic Society of America
1325 18th Street
NW
Washington DC
20036-6501
email lsa at lsadc.org

The Endangered Language Fund, Inc
c/o Doug Whalen
Department of Linguistics
Yale University
New Haven
CT 06520
email whalen at haskins.yale.edu
http://sapir.ling.yale.edu/~elf/study.html

Endangered-Languages-L Electronic Forum
c/o Mari Rhydwen
Graduate School of Education
University of Western Austrailia
Nedlands
Perth
WA 6009
Australia
email majordomo at coombs.anu.edu.au
email mrhydwen@decel.ecel.uwa.edu.au

Ethnologue
c/o Barbara Grimes
Summer Institute of Linguistics Inc
International Linguistics Centre
7500 West Camp Wisdom Road
Dallas
TX 75236
USA
http://www.sil.org/ethnologue

European Bureau of Lesser Used Languages
c/o Information Centre
rue Saint-Josse 49B
Sint-Jooststraat 49B
1030 Brussels
fax +32 2 218 1974

The Foundation for Endangered Languages
c/o Nicholas Ostler
Batheaston Villa
172 Bailbrook Lane
Bath
BA1 7AA
email nostler at chibcha.demon.co.uk
http://www.ogmios.org
was http://www.bris.ac.uk/Depts/Philosophy/CTLL/FEL

Gesellschaft für bedrohte Sprachen (Society for Endangered Languages)
c/o Hans-Jürgen Sasse
Institut für Sprachwissenschaft
Universität zu Köln
Germany
email GBS at uni-koeln.de
http://www.uni-koeln.de/gbs/e_index.html

Institute for the Preservation of the Original Languages of the Americas
c/o Executive Director
713 1/2A Canyon Road
Santa Fe
New Mexico
87501
USA
email ipola at roadrunner.com

International Clearing House for Endangered Languages (ICHEL)
c/o Kazuto Matsumura
Department of Asian and Pacific Linguistics
Institute of Cross-Cultural Studies
University of Tokyo
Hongo 7-3-1
Bunkyo-ku
Tokyo 113
Japan
email kmatsum at tooyoo.L.u-tokyo.ac.jp
http://www.tooyoo.L.u-tokyo.ac.jp

Language Documentation Urgency List
c/o Dietmar Zaefferer
Institut für Deutsche Philologie
Universität München
Schellingstr. 3
D-80799
München
Germany
email: ue303bh at sunmail.lrz-muechen.de

List Endangered-Languages-L
http://www.carmen.murdoch.edu.au/lists/endangered-languages-l/ell-websites.html

Logosphere
c/o David Dalby
Observatoire Linguistique
Hebron
Dyfed
SA34 0XT
UK
email logosphere at aol.com

Network on Endangered Languages
c/o T. Matthew Ciolek
Computer Centre
Research School of Pacific and Asian Studies
Australian National University
Canberra
Australia
email coombspapers at coombs.anu.edu.au

Society for the Study of the Indigenous Languages of the Americas
c/o Victor Golla
Department of Native American Studies
Humboldt State University
Arcata
CA 95521
USA

Terralingua: Partnership for Linguistic and Biological Diversity
c/o David Harmon
PO Box 122
Hancock
Michigan
49930-0122
USA
http://cougar.ucdavis.edu/nas/terralin/home.html

UNESCO (Study of Endangered Languages)
c/o Jean Biengen
Secretary-General
CIPSH (International Council for Philosophy and Humanistic Studies)
1 rue Miollis
75732
Paris
France
fax +33 1 406 559 480

UNESCO (World Languages Report)
c/o Paul Ortega
UNESCO Centre Basque Country
Alameda de Urquijo
60 ppal. Dcha
E-48011
Bilbao
Spain
email unescopv at eurosur.org
http://www.unescoeh.org

Universal Declaration of Linguistic Rights
c/o Follow-up Committee
Rocafort 242 bis 2n
08029
Barcelona
Spain
email dudl at linguistic-declaration.org
http://www.linguistic-declaration.org

The following organisations are mentioned in the text but not included in the Appendix:

Colombian Centre for Study of Indigenous Languages, Bogotá
Museu Paraense Emîlio Goeldi at Belém, Brazil
Academy of Mayan Languages in Guatemala
Hokkaido Ainu Cultural Centre in Sappora
Centre for Endangered Languages at Jos, Nigeria

References

[1] Crystal, David. 2000. Language Death. Cambridge, Cambridge University Press.
[2] Yamamoto, Akira Y. 1998. Retrospect and prospect on new emerging language communities. In Ostler [4] (ed.), 114.
[3] Landweer, M. Lynn. 1998. Indicators of ethnolinguistic vitality: case study of two languages - Labu and Vanimo. In Ostler [4] (ed.), 64-72.
[4] Ostler, Nicholas. ed. 1998. Endangered languages: what role for the specialist? (Proceedings of the Second FEL Conference, University of Edinburgh, 25-7 Sep 1998.) Bath: Foundation for Endangered Languages.

Initial Consultation Feedback - 3 - Individuals

Initial feedback regarding Open Echoes' strategy has also been sort from a small sample of two individuals who are foreign nationals living in London - one a national of Myanmar (Burma) regarding Burmese, the other a national of Morocco regarding Tamazight (Berber).

The feedback regarding written Tamazight was that the written form is not standardised enough to be translated outside academic institutions and another two or three years would be needed before wider participation in translation would be possible. For more information on Tamazight see http://www.geocities.com/lameens/tifinagh/

The feedback regarding Burmese concerned the difficulty in typing Burmese characters and that most uses of computers in Myanmar are reasonable English speakers. The political situation in Myanmar also makes any form of trade difficult.

Initial Consultation Feedback - 2 - Computer Aid International

Computer Aid International (http://www.computeraid.org) is the world's largest non-profit supplier of refurbished computers to schools and community organisations in the developing world and is based in North London.

Computer Aid shipped almost 12,000 PCs to partner organisations during 2003/4 financial year. Income totalled £790,000 - 75% from handling fees, 13% from grants. Major expenditure items were 30% salaries, 20% shipping.

Feedback regarding Open Echoes objectives and strategy included a recommendation to retain at least a professional project or programme management capability. Strong partnerships were also seen as essential - Computer Aid itself has strong links with the British Council, British High Commission and the UK Department for International Development. Operations based in local countries were also highlighted as key to developing donor income.

The vital issue of demand was stressed - content of whatever form must be requested or pulled rather than simply published or pushed with no demand. The politics of Open Source and commercial software in developing markets were also highlighted as complex and related Open Source projects and organisations were identified.

First Consultation Feedback

So far two organisations have provided feedback;

- Emerging Africa Infrastructure Fund - http://www.emergingafricafund.com/
- CAWD - http://www.cawd.info

The Emerging Africa Fund intends to make a real and lasting impact on the development of sub-Saharan Africa’s infrastructure to facilitate economic growth and so, both directly and indirectly over time, contribute to the alleviation of poverty in sub-Saharan Africa.

It aims to do this by focusing on the development of infrastructure and the involvement of the private sector. The fund's core product is 10-30 million US dollar denominated senior-ranking term debt with lending tenors up to a maximum of 15 years, some flexibility in loan repayment profiles and commercial (not soft) terms to finance greenfield projects with 80% equity or less equity for less risky projects.

The Charity for African Welfare and Development hosts Cawd, a network which combines networking on the Internet with community networking in rural Africa.
- Linking poor rural communities with resource-communities
- Enabling collaborations for practical problem solving
- Studying and tackling key issues regarding community development in rural Africa.

These two organisations represent the two extremes of development approaches, from large scale projects on commercial terms at one end of the scale to small scale locally based initiatives at the other.

The consultation and feedback from EAF focused on the definition of 'infrastructure' in the modern age and concluded that the current scope of the Open Echoes project does not include 'hardware' and so does not fall within the EAF's remit.

There is certainly scope for extending the project to include development funding for translation of software and then leasing of hardware and software bundles. One possible approach would be to solicit commitments from multi-nationals and governments for the supply of translated software and then finance the projects once these are in place.

The consultation and feedback from CAWD focused on the practicalities of rural African life. For example refurbished PCs do not last in the hot and dusty rural environment with intermittant power - CAWD itself has a project to provide a rugged, solar powered PC. In such poor communities expensive assets such as a PC are also expected to last 10-15 years if they are to be affordable, much like a car. This does not fit well with the current commercial operating system market which makes hardware obselete in around three to five years.

Insights into social issues were also relevant. CAWD provides training for teachers in computer awareness to prepare them for their first encounter, which may be years away. As in the world of letter writing those who are illiterate or can't access an internet cafe for cultural reasons use others to compose and review their messages before they are sent.

Nigerian graduates spend a year working in rural areas after graduation and sorely miss the internet access from their student days. They provide a source of demand for connection with the outside world and also education for local people about computers.

The historical relationship between Open Source Software and old, long-lived and esoteric equipment makes it very relevant to such poor rural environments.