Foundation for Endangered Languages

Home | Manifesto | Membership details | Proceedings | Grant Applications | Newsletter | Links | Bibliography


2. Foundation-Supported Projects

Testing of bilingual competence in children in Ethiopia Elisabeth Gfeller, SIL, POB 2576, Addis Abeba, Ethiopia


The Benishangul Region of Ethiopia is a highly multilingual area. There are 5 officially recognized local languages (Berta, Gumuz, Shinasha, Mao, Komo). A few others may still be in use (e.g. Opuo). In addition to these local nationality languages (NL), three languages of wider communication (LWC) are in use: Amharic, Arabic, Oromiffa. The resettlement policy of the former regime (for Amhara), a strong extension of the neighboring Oromo and ethnic as well as historical ties of the Berta group with the Sudan are reasons for the multiplicity of LWCs. While Mao and Komo may be threatened by their small numbers, for the other languages, the threat comes from the presence, prestige and supposedly widespread use of LWCs. The official Ethiopian language and education policy and the example of other Ethiopian languages asserting themselves and their official position provides a frame for the Benishangul-Gumuz Regional Government to develop their Nationality Languages. It seems that especially the major languages are seriously considering to pursue such a goal. The challenge to the Regional Education Authorities may be to make sure that all the five languages will be supported to assume their proper official position.

In view of eventually introducing local languages into the school system, the question arises, how all these different languages should be dealt with in the school curriculum. In Ethiopia this question is usually decided on political grounds. This study aims at adding some pedagogical criteria based on the language competence of children who enter the first cycle of primary school. Ideally, the language competence of school beginners is taken into account when language curriculum and textbooks are developed.

The argument persists, that the use of mother tongue as a medium of instruction is not really necessary, if a substantial number of children know a LWC well enough to follow the teaching of the lower primary school grades. Competence necessary to follow the school curriculum properly should include the knowledge of a certain range of vocabulary as well as the verbal ability to express certain cognitive processes. This last argument is especially relevant, if an interactive methodology is envisaged and if the home life experiences of the children are supposed to be used for building additional knowledge.

Research design

3 out of the 5 NL are planned to be developed in the next years for use in education. 30 children between the age of 5 and 8 were to be tested in each of the 3 languages. Interviewers of each language group were requested from the Regional Bureau of Education and Culture. They were given an introduction to the research instruments and then sent to 3 different villages (1 urban market town, 1 rural with no access to a bigger market, 1semi-urban with a market town accessible to the population).
As school beginners' language competence has to be assessed orally, a number of different oral research instruments were developed:
A series of cultural pictures were presented to the children who had to name them in their own language, in Amharic and in any other language they knew. It was expected that some might know at least a few words in Arabic or Oromiffa.
A series of 4 pictures were given to the children. They had to put them into a logical sequence and explain the 'story'.
A well known animal story of their own language was played to them in their language and they had to answer a few questions relating to the story. 2 of the questions asked for explicit content, 3 involved implicit information or values.
The interviewers had to fill in an answer sheet for each child they interviewed. As much as was possible to elicit, they wrote down some personal data of the interviewees and their exact answers to all of the 3 tasks.


More even than expected, methodology became a primary concern. In a first trial run, it turned out to be difficult to find a school with teachers speaking the local language. Few or no speakers of the local languages who have higher education do seem to go into the teaching profession.
Present political instability had positive as well as negative implications for the implementation of the study: government officials of one of the nationality languages who normally would work in the government structures in position of responsibility were willing to be interviewers as they were on strike for political reasons. The one female interviewer had to drop out, because she was involved in conflict resolution between two ethnic groups. The same skirmish made it necessary to change one location for interviews.

The test run also showed that the school environment was not necessarily ideal for interviewing children. A major challenge was to make them feel at ease and willing to talk. It was decided therefore to have the interviews in their homes. The village authorities however modified this request and called the children to the 'guesthouse' , which turned out to be a good solution.
Another methodological difficulty was to gather the children with the desired profile. All the children did not know their age, sometimes even older brothers and sisters or parents were not quite sure. The interviewers tended to disregard the younger children and pick those who were rather older than 8. They called the ones who looked like 5, 6 years of age 'babies' and did not expect them to be able to answer the questions. In some cases this turned out to be correct, but some younger children came up with vocabulary that the interviewers considered to be very traditional and not part of their own active vocabulary.

Due to administrative problems and the present political situation, it was not possible to run the test in all the 3 projected languages in due time (May and June 01). Only the testing of Berta children could be accomplished. It may be possible to continue with the Gumuz and Shinasha languages at the end of the year.

At least one article will be written up, when the data have been analyzed. It may be presented to a conference on 'Orality and scripturality in African languages' (organized by the Department of General Linguistics of the University of Zurich, Switzerland, October 20-22, 2001).

Discussion of results

Pictures: The 42 pictures (18 eliciting objects and animals, 18 activities and 6 feelings) were of various levels of difficulty. Children living in urban and rural areas found different objects easy or difficult to identify. A first general impression indicates that many Arabic and some Amharic and Oromiffa words are part of their current vocabulary, but only very few children used full Arabic or Amharic sentences in the tasks that elicited extended language. Many Arabic and some Amharic or Oromiffa words were used in the answers. Often the children did not discern them as being of another language, which indicates that they have become loanwords and are part of the Berta language.

Only 6 out of the 30 Berta children answered the questions systematically in more than 1 language, all the others are monolingual in Berta. It was expected that especially the urban children would use Amharic more. It seems that Amharic is only becoming the language of communication for the older children, who have been to school for more than 3 years. As in most African towns, many ethnic groups live together in their own part of town (sefer) but separated from other ethnic groups. Therefore the children use their own language for interaction with peers in the neighborhood. Only the school environment gives them the opportunity and obligation to learn an LWC.

'Misinterpretations' of pictures give some anecdotal, but nevertheless useful insights into the knowledge that the children bring to school . It is hoped that a detailed analysis of the answers will provide some indication of what cultural knowledge can be built on in the development of the Nationality Language and Social Studies curriculum.

The translation of the Berta answers into English revealed areas of vocabulary distinction that can be used in textbook development. It brought to surface distinctions that are not used in English, but may be part of the general knowledge of the age group interviewed, when they are allowed to use their own language.

Picture series: As was expected, the sequential arrangement of 4 pictures was the task that caused most difficulties. When the testing procedure was adapted and the children were first asked to explain one picture after another, a few more were able to put them in a sequence. This can be interpreted as a methodological difficulty: many children in the rural and semi-urban villages may never or not regularly see books and therefore looking at pictures may be a new process for them. The challenge for the researcher is to find tasks that could elicit the same cognitive ability using every day procedures in an oral society (giving them a few objects that could make up a logical sequence? recurring to the description of a sequential task in everyday life?)

Story: As one of the interviewers remarked, the story on the tape went too fast for the children. It was rather the adults listening at the fringe who reacted to it. To make a story digestible to children, child's language would need to be used. One child, who was asked to give a summary of the story, used what the interviewer called child language: he pointed out a recurring syllable as characteristic for child speech (discourse feature marking reference?). This feature may need more linguistic research as it may have important implications for the type of language to be used in textbooks for early grades.


As this study is one part of a long term project aiming at providing linguistic and cultural data necessary to provide a culturally adapted nationality language curriculum, overall conclusions cannot yet be drawn. Preliminary conclusions drawn from the bilingual competence testing with Berta children affect the following areas: research methodology in societies with basically oral communication patterns, language competence and cultural knowledge of school beginners.

Research methodology in basically oral societies. It is completely 'unnatural' to isolate a single child for testing. People - young and old - passing by were part of the testing scene in all three villages. Especially the children in the lower age range often needed the support of a familiar sibling before they even uttered a single word. It may be useful to try out some kind of group testing procedures. An interviewer interacting with a group of children and one or two observers keeping track of the answers of individuals may make group testing possible and allow to get some idea of differences between the performance of distinct individuals.



Fortunately most interviewers were surprisingly well adapted to deal with children. The research implementation asked for adaptation to actual circumstances, the respective local situation and cultural implicits unknown to the researcher. With interviewers who were quickly grasping the research objectives and the most important aspects of the methodological approach, it was possible to elicit useful data in rather difficult circumstances.

Language competence. Final conclusions have to wait for a full analysis of the data. At this point, the hypothesis is confirmed, that school beginners are more apt to express themselves in their own language at a level necessary to interact with curriculum content than with an LWC (including Amharic as well as Arabic and even more so for Oromiffa). The surprising finding is that few urban children in the tested age group were sufficiently fluent in the LWC to use it efficiently for subject teaching. A thorough introduction of any LWC is indicated, even teaching it as a foreign language, at least for the first few school years.

The indication that Berta is clearly a dominant language for all the children interviewed (except 1 or 2 balanced bi- or trilingual individuals) may ease the often difficult question of first language choice in a multilingual environment.

Cultural knowledge. As the present study did not explicitly aim at cultural findings, only a few provisional statements can be made regarding this topic. As language and culture are closely linked and linguistic as well as cultural aspects are crucial for effective teaching in lower grades it can be expected that linguistic data will provide some cultural output and vice versa. Comparing pictures that were easy to interpret for many or all children, versus pictures that were interpreted in many different ways can will give some pedagogical insights for the production of child conform textbooks - referring to content as well as to language.


I would like to thank the Foundation of Endangered Languages for their financial support for this research. The contribution allowed me to give some financial remuneration to the interviewers who had to learn quite unexpected skills and who had to use patience with the young children, something they may not usually be asked to do in their normal jobs.

Khang -an endangered language in Vietnam
Sent: Friday, June 08, 2001 9:01 AM
Prof. Dr. Nguyen Van Loi
Institute of Linguistics, Hanoi, Vietnam

The Khang (also called Makang, Sa Khao) are an ethnic group with a population of 3,821 people, living in the Son La, Lai Chau, Lao Cai provinces of North-Western Vietnam. The Khang language belongs to North Mon-Khmer division of the Austroasiatic family.

Khang consists of some dialects and subdialects. It has accepted many changes in its vocabulary (borowing from Black Thai words), phonetics (the presence of tones), and grammar (the absence of prefixes, infixes...), resulting from contact with Thai over many centuries.

Khang is spoken by oldest, the young now mostly speaking in Vietnamese, and Black Thai. As a result, Khang is an endangered language.

The documentation, primary study and description of Khang language is the aim of our small project with the grant from the Foundation for Endangered Languages. Up to now our project achieved as a result of the grant we received the following:

1. Field work to collect the documents on Khang dialect in Ban Gion village, Muong La district, Son La province. (May 18 to June 5, 2000)
· Prof.Dr. Nguyen Van Loi, Institute of Linguistics, Vietnam.
· Dr. Ta Van Thong, Institute of Linguistics, Vietnam
· Dr. Nguyen Huu Hoanh, Institute of Linguistic, Vietnam
Documents are including:
· Basic vocabulary (1000 words)
· 100 sentences indicate some basic features of grammar of Khang language.
The data were transcribed in IPA and recorded by digital audio tapes (DAT Sony)

2. Storing data, collected in Khang Son La dialect in computer and CD ROM. Data have been found in "The Database of Endangered Languages in Vietnam" of Institute of Linguistics, Hanoi, Vietnam.

3. Analysing data, collected in Khang Son La dialect by program ASAP (Acoustic Speech Analyse Program) and CECIL (Computerized Extraction of component of Intonation in Language).

4. Comparing analysed data of Khang dialects: Khang Son La, Khang Quang Lam, Khang Lao Cai

5. Describing the phonetic system of Khang Son La dialect.

6. Describing the tonal systems of Khang language (Dialects: Son La, Quang Lam, Lao Cai )

7. Articles written:
The position of Khang in genetic classification of languages in South-East Asia. (In Vietnamese, Published in Journal Ngon Ngu, 10 (2000)
The Khang Language in Vietnam. (In English, Will appear in Languages on the Chinese-Vietnamese Border ed. Jerold Edmondson.)

Lavukaleve dictionary project

Angela Terrill , Max Planck Institute for Evolutionary Anthropology

The Lavukaleve dictionary project began in 1995, when I started working on producing a grammar of Lavukaleve, an indigenous language of the Russell Islands, in the Solomon Islands. When I started my work there I was asked to compile a Lavukaleve-English dictionary to help preserve the language and help children in school learn English. The dictionary work was started with funding from the Australian National University, and was enabled to be continued on the basis of funding from the Foundation for Endangered Languages.

There are about 2,000 speakers of Lavukaleve; of these, very few people use written Lavukaleve. Indeed few people have a need to write at all. The written word is mostly confined to a few liturgical services which have been translated into Lavukaleve. Official English-only language policy in Solomon Islands schools means that even those who have attended school have not been taught how to read or write Lavukaleve. However, Lavukaleve is phonologically relatively well suited to an alphabetic orthography, and there seem to be few difficulties for people in reading and writing Lavukaleve words when necessary.

Given this literacy context, what role would the dictionary have? It was to be made for people who do not have a reading culture, and who were already native speakers of their language. I discussed with community leaders and school teachers what form the dictionary should take. Opinion was fixed that it should be a bilingual Lavukaleve and English dictionary. I asked whether Solomon Island Pijin should be included, given that more people know Pijin than English. However, people pointed out that Pijin is not considered a written language, no-one knows how to write it, and in any case it is not considered a suitable language for such a purpose. In Solomon Islands, Pijin is, perhaps unfortunately, denigrated as being a corrupted variety of English, and thus nobody wanted Pijin in the dictionary.

Once the languages for the dictionary were determined, there were still questions about what it should contain. The dictionary was intended for Lavukaleve speakers, but realistically, it is also the only dictionary ever likely to be made on this language. Therefore it should be useful and useable by Lavukaleve speakers, but it should also be useful for other non-Lavukaleve speakers who want to know about the language. In the future, government policy might change and the dictionary could conceivably be the basis for pedagogical materials. Other linguists may find it useful in the future, so it should have grammatical information and example sentences. However it should not have so much grammatical information as to be uninterpretable and off-putting for the very people it was intended for.

So far the Lavukaleve dictionary is in its second version. It consists of 160 pages, starting with an introduction to how to use the dictionary, the Lavukaleve-English section which comprises the main body, and an English-Lavukaleve finder-list. It is co-authored by myself and my main consultant, Patterson Barua. It is a compromise between the desires of the community, and possible future needs of the language. It contains, as well as words and their translations, example sentences for verbs and adverbs; information on word class and information on transitivity for verbs, and for nouns, their gender, their dual and plural formatives and irregular possessive and locative formatives. There is also a small amount of encyclopaedic information, particularly with respect to plant names, explaining what the plant looks like and what it is used for. It is an ongoing process; every trip I find out more information and correct mistakes.

An underlying question remains, why do people who don’t read want a dictionary anyway? The answer to this question was made apparent to me very early on when I first asked what work Lavukals wanted me to do on their language. Most people answered that they wanted a book written on their language just like that on one or other of their neighbouring languages. Specific comparisons were made to their own language, which had never had any outsider work on it, with those of many other languages that had. A picture emerged of their sense of their own language as a worthless language, not deserving of respect; if it had been worthy, it would surely have had a book written on it. There was a widespread perception that the Lavukals, and their language, had been forgotten by the outside world.

Having a book written specifically on a community’s language is a matter of great pride for that community. To this end, the content of the Lavukaleve dictionary is never the matter of comparison with those of other languages, but rather the size and substantiality of the book is. This is not a trivial observation, but rather is an overt manifestation of a deep unease about the worth of their language in comparison with other “outside” languages, and, ultimately, about the growing encroachment of urbanisation and globalisation and its impact on the indigenous culture and way of life. Within this context, the value of the Lavukaleve dictionary for this community has less to do with its content, and more to do with its very existence.