Volkswagen Foundation Programme

Documentation of Endangered Languages

Project proposal (main phase, August 2001)

The EGA documentation model:

Evaluation and application to selected West African languages

1. Project partners

Proposer / coordinator:

Project scientists:

1. Abstract of main phase extension proposal

The main phase of this project has as its foremost goal to build on, scale up and evaluate the results of the pilot phase by applying the methodology developed during the pilot phase to a selected set of typologically and sociolinguistically different endangered languages in West Africa. In the following sections, we summarise the project methodology, the project structure, and the motivation for extending the domain of the main phase project in comparison with the pilot phase project.

Summary of project methodology

The methodology developed in the pilot phase can be characterised by the following features:

1. Integrated approach to multi-level language documentation.

2. Multimodal empirical base: digital audio, with parallel laryngograph and airflow signals.

3. Standardisation of resources: fully explicit documentation of methodology (reports, manuals), use of and contribution to international standards (XML and metadata conventions), cooperation with international initiatives (EAGLES/ISLE, COCOSDA).

4. Evaluation of results in order to ensure scalability beyond single languages.

5. Star multiplier strategy: intensive scientific and practical cooperation with local and regional institutions and embedding of the results in heritage archiving.

6. Feedback to the local speaker community by active involvement in documentation, training young people, provision of language primers.

The project documentation methodology is firmly rooted in scientific interest, in particular interest in the contribution of little-known and endangered African languages to the advancement of theoretical and applied science in several disciplines in the humanties, by efficiently producing documentation of a language and culture according to a Principle of Explicit Documentation (see main part of proposal).

Summary of project structure

The structure of the project has a certain complexity which marks it (1) as an infrastructural effort, with multiplier function on the basis of intensive local cooperation, (2) as a design project, in which computational linguistic tools for efficient documentation are designed and given a "proof of concept" implementation as input to the main database team, and (3) as a documentation programme in which a selected group of languages with different scientific interest and different - though ultimately related - endangerment types are efficiently documented using the pilot phase methodology.

The main phase of the project is planned for four years, in two two-year phases. In addition to ongoing empirical documentation activity in each phase, each phase has a somewhat different focus:

Project Phase I: upscaling the methodology by further systematising collation, description and archiving procedures, for the production of manuals, and for training workshops and evaluation of existing procedures and materials.

Project Phase II: consolidating and completion of the documentation, and publication with three main goals in mind: creation of materials for the local community; coooperation with regional archives; dissemination of materials and documentation through internationally archiving centres, in particular the MPI endangered language archive in Nijmegen.

Summary of motvation for broader documentation domain

The specific features of the languages concerned shed light on the cognitive range of the human language faculty: extending the known kinds of sound change (tone, unusual consonant types) and of types of lexical item, as well as of the culturally functional diversity of language in context (trade, taboo areas, music, art).

1. The project methodology iw well worked out and appears suitable for more extensive application.

2. Close institutional contacts exist with West African universities, in particular in Ivory Coast and Nigeria, which include joint publications with local scholars, cooperation with former Humboldt scholars (Ahoua, Urua) who have experience of modern language documentation techniques, and involvement with the West Africal Linguistics Society (Connell is a member of Council , Urua is a former election officer, Gibbon is responsible for the WALS website).

3. Africa is known to be by far the most linguistically diverse continent. The Niger-Congo phylum, in particular, is well known for its linguistic diversity, with >1400 languages, >70 mutually unintelligible Niger-Congo languages in Ivory Coast, >400 in Nigeria. In particular the unique features of West African languages have long been well known for the crucial contribution which study of their specific properties has made to linguistic theory and the understanding of linguistic and cultural change. This extreme diversity cries out for upscaling present methodologies.

4. By strategic use of the 'star ultiplier' strategy of `trainees as trainers', which has been used effectively in the pilot phase, we aim for added value in language documentation beyond the project itself. Although the specific contribution of the project can be no more than a drop in the ocean of language diversity and language endangerment, concentration on developing, evaluating and spreading the use of an efficient empirically and theoretically well-founded methodology in cooperation with local institutions ensures maximal impact of the funding programme.

5. Careful precautions have been taken during planning to reduce possible centrifugal forces involved in broading the scope. Excellent existing cooperation between all participants, exchange of materials, and the organisation of annual regional workshops are designed to support the coherence of the project. A 'staggered' structure is designed for maximally economical use of human and material resources in the partner institutions, with an initial two-year phase essentially for training, testing and upscaling the methodology, and a further two-year phase which is essentially for consolidating the methodology.

Summary of operational objectives

The project aims to produce concrete, operational results of the following three kinds:

1. Multitier multimodal documentation of five West African languages, with supplementary documentation in the form of detailed phonetic descriptions, a 1000-2000 word lexicon, sketch grammar and morphology, documentation of language usage in context with video footage, sociolinguistic questionnaire results on locating the languages in an endangerment space defined by a well-defined set of endangerment factors.

2. User designed computational linguistic tool prototypes, particularly for computational lexicography with concordancing and other corpus processing functions.

3. Training of local and regional personnel to design and perform documentation expeditions with modern equipment, and multitier multimodal documentation, by means of local and regional workshops under the aegis of the West African Linguistics Society, and by means of short-term training scholarships.

4. Development of constructive working relations with the speaker communities, in particular by provision of orthographic primers and reading material in the languages in cooperation with the village administration, and by training parents in the community in their use.

Judging by preliminary results achieved in the pilot phase, these goals, though ambitious, are achievable with the personal and institutional cooperative structures which have been put in place for the project main phase.

1. Detailed project description

A. Overview of proposal

The present section covers a brief report of the pilot phase results (the main report is due in September 2001), a detailed description of the proposed main phase project in terms of the languages to be described, the documentation methodology to be followed, and the local cooperation and training strategies.

B. Overview of pilot phase goals

The pilot phase was conceived as a model documentation pilot project in the initial phase of the funding programme, focussing on one language, Ega, within a frame of reference of data orientation, multifunctionality and general accessibility. These general operational goals were supplemented with the additional objective of building local infrastructure for documentation. A major intellectual and scientific goal was to understand factors in a model of language endangerment space, as well as to ensure that strong scientific interest in the typology of the language itself was catered for.

C. Overview of pilot phase results

All the pilot phase objectives were achieved. Detailed accounts of the results accompany this proposal in the form of project reports and on CD-ROM, and will be detailed in the pilot phase final report which is due at the end of September 2001. We summarise here the results of the main dimensions of work: the Ega project methodology; the notion of endangerment space from a sociolinguistic point of view; linguistic and phonetic documentation results; special phonetic documentation; specific types of linguistic documentation; multi-tier annotation; computational linguistic prototype development; local training infrastructure and institutional cooperation; general operational results.

The Ega project methodology.

The pilot project has concentrated on Ega, a language that fulfils accepted criteria of endangerment, strong scientific interest, and openness of the population to practical feedback. The project partners in Bielefeld, Canada and Abidjan have developed a methodology with the following features:

1. An integrated approach to multi-level language documentation which takes up and contributes to the relatively new paradigm of language documentation which has been developed in parallel at several centres world-wide since the early 1990s, to a considerable extent under the influence of speech engineering and computational corpus linguistics, and strongly supported by the project funding by the European Commission in the area of linguistic resources for language documentation and technology.

2. In view of the theoretical and practical impossibility of interpretative analyses when languages have become extinct, in addition to documentation of language in use, considerable effort has been devoted to providing the highest possible quality of signal-oriented (phonetic and phonological) annotation for audio indexing and later linguistic analysis.

3. Standardisation of resources, with fully explicit documentation of methodology (reports, manuals), and extensive use of and contributions to international standards (phonetic transcription formats, annotation lattice formats, XML based document description languages and metadata conventions). This dimension of the methodology is supported by intensive cooperation with international documentation and standardisation initiatives (including EAGLES, ISLE, Talkbank, COCOSDA), and by the commitment to provide project results to the central archive.

4. Evaluation of results in order to ensure scalability beyond single languages. This is standard procedure in system and documentation development, but not immediately obvious in the context of traditional linguistic and anthropological descriptions, particularly where linguistically relativistic approaches are in the forefront, and the methodology of investigation itself is not strongly focussed on. We consider this condition to be a sine qua non of good documentation.

5. Star multiplier strategy: in order to maximise the impact of the initiative for documenting endangered languages, intensive scientific and practical cooperation with local and regional institutions with a view to embedding the results in regional heritage archives has been established.

6. Close contact with and feedback to the local speaker community by active involvement in documentation, the provision of language primers and lexica, and training young people in their use.

The project documentation methodology is firmly rooted in scientific interest, in particular a strong interest in the contribution of little-known and endangered African languages to the advancement of knowledge in theoretical and applied science within several disciplines in the humanities. In the interest of efficiently producing documentation of a language and culture for posterity, we proceed according to a Principle of Explicit Documentation:

Language documentation is explicit if it enables insightful study of a langauge and culture when the language is extinct and when native speakers are no longer available as sources of creative, operational, structural and interpretative knowledge about the language.

We consider this principle essential. It has consequences both in terms of the highest possible quality of documentation at levels close to the speech signal (phonetics, phonology), and in terms of detailed interpretative glossing and commenting. The principle is, of course, implicit in the urgency and comprehensiveness criteria of the VWS funding documentation; for ourselves it is an essential criterion for ensuring that our work is based on solidly justifiable methodological criteria, over and above our sympathy for and fascination with cultural differences. The methodology developed in the pilot phase is described in detail in the following main section of the proposal.

A simplified overview of the logistical strucure involved in the Ega project methodology is shown in Figure 1.

Preliminary results: "endangerment space"

One of the main results was recognition of the need to define a notion of endangerment space within which the functional and structural "degrading" of a language can be orientated along sociolinguistic structural dimensions. In a sense, the whole documentation is relevant for this notion, but the project concentrated on sociolinguistic work on village profiles.

Questionnaires investigating language knowledge and use have been administered in various locations in the Ega region. These were designed, first, to ascertain the extent to which Ega is actually spoken and, second, the range of facts and factors influencing the use of Ega. One questionnaire is termed the 'Village Profile' and seeks to elicit general information about the village being investigated. The other questionnaire is a form of 'Linguistic Census', administered on an individual basis, and seeks to elicit specific information as to which language is used in particular situations, as well as information concerning attitudes to language use. Both questionnaires have been conducted in five Ega villages located in different parts of the Ega region. In total there have been several hundred respondents to the linguistic census. Attention should be drawn to a general methodological aspect regarding the use of such questionnaires. There is a recognized concern that 'self-reporting', such as is required in the linguistic census may not reflect the true situation, and a follow-up involving a small subset of the respondents is recommended. We have overcome this in using two Ega assistants (one a student of sociology) in conducting the surveys, each an indigene of a different village in the set of surveyed villages and highly fluent in Ega and French, both of them well known to the respondents, and able to verify the information provided.

The results of these surveys have been entered into a database and are available for further analyses. Preliminary analysis of the five Village Profiles and the Linguistic Census from one village have been completed and the results presented at the 32nd ACAL at the University of California, Berkeley, in March 2001, and at an invited seminar in the Dept of Languagess, Literatures, and Linguistics, York University, Toronto, in June 2001.

It is often held that language endangerment is "normal" as a special case of language change, and that it may be associated with language birth via re-lexification from a dominant language - except that increasingly today language births are being greatly outstripped by language deaths, Ega being no exception in this respect. The surveys have confirmed, in particular, that that a language is endangered when its functionality is degraded (independently of its linguistic properties), and that this happens when the speaker community is small, the language is used for fewer and fewer public functions, finally remaining only in ritual and in the language of the elderly at home. In the case of Ega, this can be clearly seen in a number of peripheral villages, while the core villages have comprehensive language functionality.

But an entirely different problem from this classic description of the symptoms of language endangerement is the question of the aetiology of language endangerment. For Ega, a number of putative causal factors have been identified (though, experimental procedures being inapplicable, not in a strict quantitative sense):

1. Many mothers speak the dominant language (Dida, a Kru language) because of the exogamous marriage conventions which are widely adhered to.

2. The male population migrates in search of work.

3. Males from other ethnic groups (Dioula, Baule, Mosse) migrate into the area looking for work on the plantations.

4. An important factor in these migrations is climate, involving globally influenced climate changes as well as regional deforestation of tropical rain forests during the past century and their replacement by cocoa, coffee and tropical fruit plantations, and wide-area flooding for reservoir construction.

5. With the migrants, an attractive migrant language such as Dioula arrives, which enables fluent communication with traders from other regions (in some villages with highly mixed ethnicity, Dioula has become the playground lingua franca of schoolchildren).

6. Finally, a repressive language policy contributes to endangerment, as when schoolchildren are forbidden to use the language in school or even in the playground.

As far as the history of the Ega people themselves are concerned, results are somewhat inconclusive. Village history varies wuite widely from village to village and appears to have arbitrary elements (in some villages, the claim is that the Ega came from the North-East, from Liberia, in others that the Ega came from the East through neighbouring areas occupied by the Abbey (with whom a long standing cooperation pact exists), in others simply "from across the river", or from a neighbouring large town, or "have always been here". There are folk etymologies for "Ega" with other ethnic names such as "Ga", and even "Gha"-na; Ghana of course did not exist at the time when Ega migration is thought to have taken place, but a grain of historical fact may be present in that Ega migration was indeed almost certainly westwards, though whether by sea or by land or both is not known. This may indicate, perhaps, that the group has been settled for so long that historical change is not an issue, and is arbitrarily reconstructed ad hoc on demand. There is a large stock of lexical item and idiom borrowings from the enclaving Kru language, Dida, and only a fairly small common stock of words between Ega and the Kwa languages, even the geographically nearest ones, which are located at some distance to the East, although Ega has so far been tentatively classified as Kwa. On the other hand, there appear to be some striking commonalities with the core lexicon of Benue-Congo languages which have so far not been documented or analysed in detail. Some aspects of cultural life, such as the use of "drag nets" for hunting, may point to a history of the Ega as a lagoon, fishing or sea-faring people, though there is apparently no recollection in oral history of anything like this, and cultural borrowing (though of a particularly bizarre type) may be the explanation.

Preliminary linguistic and phonetic documentation results:

In the case of Ega, loss of the language without documentation would lead to an irreparable gap in our knowledge of the possible complexity of sound systems and language universals, and of language change in the context of the migration of ethnic groups. We have therefore concentrated on these aspects.

Three main kinds of data were elicited: (1) audio and video recordings of language in its context of use; (2) questionnaire directed elicitation of systematic morphological and grammatical paradigm data; (3) data directed by experimental phonetic criteria. Lexical data have been abstracted from all three types. These three types were combined in order to get an overview both of language form and structure at different levels of granularity on the one hand, and language functionality on the other. The epistemological distinction between form oriented properties of langauge and interpretative or hermeneutic roles of language has often been seen as a source of fundamental controversy and indeed ideological warfare between different methodological factions in empirical linguistics. We do not accept this incompatibility assumption, but consider the distinctions to be essential and complementary. On the one hand, we consider it to be common knowledge that exhaustive information about the forms and structures of a language or indeed its meanings and uses, cannot be acquired from corpus collation and analysis alone, both on a priori grounds (the creativity of language), and on a posteriori grounds: in a given corpus, there are large gaps in the range of forms which occur, for many reasons. Explicit questionnaire-based paradigm elicitation procedures are necessary part of the methodology, and are not simply a contribution to the efficiency of data elicitation. On the other hand, traditional varieties of elicitation for typological purposes give a meagre picture of the language in its cultural context in ignoring the ecology of language data; interpretative criteria are a sine qua non in documentation, languages not being systems of perceivable patterns alone, insightful though these patterns may be in regard to the cognitive potential of human beings.

Phonetic documentation

In view of the paucity of available information, and the great scientific interest in the unusual properties of the sound systems of African langauges, systematic experimental phonetic data has been recorded for the investigation of the Ega sound system. To achieve this, speech materials (wordlists) were compiled exemplifying all phonemic contrasts in the language: viz. its system of nine vowels, featuring [ATR] harmony; its consonant system with 27 contrasts which includes, almost uniquely in the world, distinctive implosives at five places of articulation, as well as an interesting set of depressor consonants; and the tonal contrasts and combinations generated by its three tones. Multiple repetitions of these materials were recorded by 16 speakers (six for the tone material), using a stereo digital recorder with the audio feeding into one channel and a laryngograph (electroglottograph) feeding into the other channel. One set of the speech materials has been marked and labelled for illustrative purposes in the documentation. The entire set is ready for detailed phonetic analyses. In addition, a subset of these materials (viz. consonants participating in the implosive - non-implosive contrast) was selected for investigation via airflow and pressure recordings.

From the point of view of documentation methodology, one aspect of the interest of the laryngograph and airflow recordings is their multimodal character: they involve parallel channels of information about the speech signal, which require synchronisation in order to be archived and used at a later date. The problems of this type of documentation, which are partly analogous to the problems involved in the parellel use of video and audio, are being discussed closely with the central archiving team (the TIDEL project).

Linguistic documentation levels

A number of linguistic studies were prepared, partly with graduate students in Abidjan, partly by colleagues with task contract consultancies in Abidjan. These include:

• lexical database with glosses and grammatical information (currently ca. 1000 words from various sources)

• questionnaire-based sketch morphology and grammar for simple sentences and complex sentences in discourse contexts

• detailed phonological description with phonetic realisation rules

• phonemically based orthography (and an introduction to its use for native speakers)

Multi-tier annotation:

The main focus of the documentation is on annotation of the speech signal according to the agreed conventions of the pilot phase projects, and following international standards. The specific results of this activity include recordings of the questionnaires and phonetically guided recordings, but most centrally:

• a preliminary collection of ceremonial narratives by the head narrator of the Gniguedougou village and his assistants (three with video documentation)

• transcription of all narratives in IPA (in standard SAMPA ASCII typographic conventions)

• detailed multitier annotation of one medium-length narrative using standard software (Praat, Transcriber), both in Abidjan and in Bielefeld (for quality control by cross-checking).

Synchronisation of the annotations with multi-modal parallel signal channels is being conducted in cooperation with the database tool project.

Computational linguistic prototype development

One of the specific features of the project is the strong component of experience and qualification in computational linguistics, particularly the computational lexicography of spoken language and its neighbouring subdisciplines of computational phonetics, computational phonology, and computational morphology. This component of the project produced as its main result a "proof of concept" research prototype of a concordance for spoken language corpora (PAX: Portable Audio Concordance System), which automatically generates a concordance (with enhanced output options including waveform, spectrum and pitch visualisation) from a SAMPA annotated corpus. This prototype is being used as it stands (and is already available as an interactive Internet application), but is fundamentally intended as a research design study to provide operationalised specifications for end-user oriented implementation by specialist programmers.

In addition to the lexicographic work, a family of software filters was developed in order to guarantee compatibility between annotation systems which use different software; so far, filters which inter-convert Praat, Transcriber, esps/waves+ annotation formats into a standardised XML format (TASX - Time Alignment of Signals in XML). This was an interesting problem rather than just a practical obstacle, in view of the different temporal logics which implicitly underlie the different annotation formats. The approach taken was influenced by the proposer's previous work on standardisation of formats, and by the Bird/Liberman annotation graph approach (see discussion in the pilot phase proposal).

The role of the computational lexicographic work on the audio concordance are shown in Figure 2. The initial empirical base documentation is designed and produced by the linguistic or anthropolgical fieldworker, yielding a multi-tier annotation. The PAX software takes this as input, and yields an interactive concordance as output, in three stages. First, tier-specific wordlists are automatically extracted from the annotation files. Second, the user queries PAX using the wordlist and other options, and receives transcriptions of the contexts in which the words occur (the basic KWIC - Keyword In Context - technique), plus options for selecting an audio replay, and waveform, spectrum and pitch visualisations.

Local training infrastructure and institutional cooperation:

In a number of visits to Abidjan and Bielefeld by project personnel from the partner institutions, intensive training in language documentation was conducted. A highlight in this training programme was a workshop in February 2001 in the Computer Science section of IRMA (Institut de Recherche en Mathématique Appliquée) at Université de Cocody, Abidjan, which was attended by 20 colleagues and graduate students, and laid the foundation for further detailed documentation work, not only on Ega but also on the languages being worked on currently in connection with doctoral theses in other areas.

General operational results:

The general operational goals for the pilot phase project which were outlined above have been reached in the following ways:

1. Data orientation: A basic data set of (a) digital audio (and short video) recordings of different genres (in the broad sense of `linguistic data type'), from `ecologically valid' story-telling and everyday dialogue to systematic experimental phonetic field data, (b) linguistic survey material, (c) sociolinguistic survey material on endangerment factors, was designed during the first 4 months of the project with intensive internet based cooperation with local personnel. In February and March 2001, Ahoua, Connell and Gibbon spent a total of 3 person-months in Abidjan (training seminars for local staff) and in Ega country (data elicitation for recordings and surveys). Different selections of this material are currently being annotated by project partners. A core lexicon and corpus concordance have been prepared.

2. Multifunctionality: Care was taken in designing the corpus to ensure that the data selection and outline analyses (`sketch grammars', lexicon etc.) would be of interest to linguists, phoneticians, specialists in oral literature, anthropologists, as well as educationalists and language planners working on alphabetisation materials.

3. General accessibility: The data results are all available in common hypertext and multimedia publishing formats; an XML format, suitable for open architecture language documentation workbenches, has been developed in cooperation with colleagues inside and outside the DOBES consortium. Software tools for conversion between specific annotation formats (Praat, Transcriber, esps/waves+, XML) were also developed and are being made generally available. Scientific results are immediately made available on the project website and at relevant scientific conferences.

4. Local infrastructure: Establishment of local infrastructure on the `star multiplier effect' principle (locally trained personnel themselves train others) has been established at a number of levels: (a) Dr Ahoua, Abidjan, is a project scientist of the project; (b) colleagues in Abidjan have been involved with small consultancies for contributing to the lexicon and sketch descriptions, the computational infrastructure and concordance construction were supported by computer scientists at IRMA (Institut de Recherche en Mathématique), Abidjan (see also co-published reports); (c) a doctoral student from Abidjan was employed as consultant in Bielefeld for designing annotation criteria; (d) partners were invited to advise the Ministry of Culture and Communication on modern documentation techniques; (e) working contacts were established with INSAAC, Abidjan, the College of African Art and Culture, for anthropological aspects of the main phase.

Documentation of the results:

The details of these results are documented in four main ways:

• Project reports: submitted with this proposal.

• Conference and workshop presentations: (1) Presentation at West African Linguistics Conference August 2000, Accra, Ghana; (2) Presentation at Colloquium on African Languages and Linguistics, August 2000, Leiden, NL; (3) Presentation at LINGUISTIC EXPLORATION: Workshop on Web-Based Language Documentation and Description, December 2000, Philadelphia, USA; (4) Presentation at Annual Conference on African Linguistics, March 2001, Berkeley, USA; (5) Invited presentation at EAGLES/ISLE Computational Lexicon Working Group Workshop, March 2001, Pisa, Italy; (6) Presentation at Typology of African Prosodic Systems, May 2001, Bielefeld, Germany; (7) Invited plenary address at Poznan Linguistics Meeting, May 2001.

• Project web site: http://www.spectrum.uni-bielefeld.de/LangDoc/EGA/ containing project reports, format specifications, sample data, on-line `proof of concept' concordance implementation.

• CD-ROM: submitted with this proposal.

Publications based on the reports and presentations are in progress. A comprehensive report on the specific model documentation goals will be presented at the end of the pilot project year (end of September 2001).

D. Abstract of main objectives for main phase extension

Initial evaluation and testing of the methodology will take place on two further language in Ivory Coast, Eotile (a Kwa language, not closely related to Ega) and Mbre (presumed Niger-Congo, but precise affiliation unknown), both of which are highly endangered, but on different grounds to Ega. Parallel to this, the methodology will be gradually re-applied in cooperation with colleagues at the universities of Uyo and Port Harcourt to two other Niger-Congo languages, Defaka and Nkoroo, both of which count as highly endangered mainly because of the oil economy of the Niger delta, but with Defaka (which is enclaved by Nkoroo) in the greatest danger of extinction.

Considerable interest has been aroused by the goals and methodology of the project in linguistic departments at various West African univerities, with the vision that modern methodologies of this type could serve to support existing work on language documentation, particularly of endangered languages, in the context of cooperation between West African linguistic centres in both francophone and anglophone areas. The idea was born, mainly in discussions with Prof. Eno-Abasi Urua of University of Uyo, Nigeria, who spent the academic year 2000/2001 at University of Bielefeld, that the planned extension of the methodology should therefore cross post-colonial national borders and take into account highly endangered Nigerian languages. The two languages selected by Dr Connell in discussion with Prof Williamson and Dr Blench are Defaka and Nkoroo, two Ijoid languages, both endangered, but with Defaka additionally enclaved and threatened by Nkoroo.Both are of high scientific interest in regard to the reconstruction of the interrelations between the Niger-Congo languages. Dr Connell estimates that this extension is not only a feasible and valuable step towards efficiently documenting more endangered languages, but also a highly desirable infrastructural move.

Another factor in designing the main phase extension is that the extinction rate of threatened languages turns out to be higher than expected.

Finally, it appears to the proposers that concentration on singular documentation efforts for particular languages, while of great value in themselves, does not justify the large amount of effort involved, and that therefore evaluation, upscaling and generalisation to other languages in other contexts is a necessary scientific and logistic goal.

Accordingly, the scope of the main phase of this project, with its foremost goal the evaluation, testing and re-application of the methodology developed in the pilot phase in order to estimate its scalability in a strictly selected set of broader language documentation contexts, has been somewhat broadened, though in a way which is strictly constrained by the given criteria.

There are two aspects to the broadening of scope. First, initial evaluation and testing of the methodology will continue in depth with Ega, but will also take place on two further language in Ivory Coast, Eotile (a Kwa language, not closely related to Ega) and Mbre (presumed Niger-Congo, but precise affiliation unknown), both of which are highly endangered, though on different grounds to Ega (see below). Slightly later, the methodology will be applied to the two selected Nigerian Niger-Congo languages, Defaka and Nkoroo. Both of these languages count as highly endangered, partly because of the dominating oil economy of the Niger delta, but with Defaka in the greatest danger of extinction. We assume that the notion of endangerment space developed in the Ivory Coast work will also be applicable, mutatis mutandis, to these Ijoid languages, and that there will be clearly recognisable linguistic, sociolinguistic and cultural endangerment factors. This is an issue of the typology of endangerment, and does not necessarily imply direct influence or recent common history via traditional West African land and sea migration patterns.

The Nigeria arm of the extension has been nplanned in cooperation with colleagues at the near-neighbour Universities of Uyo (Dr Eno-Abasi Urua) and Port Harcourt (Prof Kay Williamson) in Nigeria, and in consultation with a leading specialist on Niger-Congo languages, Dr Roger Blench (London). One of the project proposers, Dr Connell, has spent many years (part of this on the staff of the University of Calabar, Nigeria) working on endangered languages in the Cameroon-Nigeria borderland, and is in an optimal position to supervise the scaling of the methodology on the Nigerian side.

The partners are aware of the coordination challenges facing them in the implementation of this project design, and are confident that their previous work, both prior to the Ega pilot project work, and within the pilot project, has provided a scalable methodology for successful efficient language documentation. We are also confident that the project will be fully supported by local personnel at the cooperating universities.

Table 1 summarises the structure and activities proposed for the main phase (see also the detailed task and time line descriptions below).

E. Côte d'Ivoire-Nigeria coordination measures

Coordination of project work between Ivory Coast and Nigeria requires only a small extension of the email and web-based internet telecooperation techniques which have been successfully used in the pilot phase. All project partners have known each other for many years. Two of the partners, Dr Ahoua (Ivory Coast) and Dr Urua (Nigeria) have spent considerable time in recent years as Humboldt scholars at Universität Bielefeld, during which time they gained, among other things, experience in the fundamentals of language documentation.

Both Prof. Gibbon and Dr Connell have excellent working relations with both centres. Prof. Gibbon has worked with colleagues in Ivory Coast for some 18 years, and Dr Connell has worked in Cameroon and Nigeria on endangered languages in these countries for the same number of years.

The following specific coordination measures are planned:

1. Each year, the partners will hold a coordinating in loco report and training workshop, in principle alternately in Abidjan, Ivory Coast, and Uyo, Nigeria. Additionally, Symposia are planned independently of but in temporal juxtaposition to the West African Linguistics Congress 2002 and at the following WALC 2004. The workshops will take place with Dr Jaques Rongier, Dr Roger Blench and Prof. Kay Williamson, as external advisers to the project, and with invitation of further internationally reputed experts on language endangerment in Africa and on language documentation. Funding for these Symposia is applied for.

2. As far as possible, the partners will all be involved in fieldwork and general training activities both in Ivory Coast and Nigeria.

3. The partners will exchange sample materials of various documentation types on the basis of the standards developed in the DOBES consortium.

4. The partners will continue to acquire additional funding for doctoral students as a multiplier effect of this project.

F. Motivation for the Côte d'Ivoire branch of the extension

We estimate that approximately 24 months of annotation and sketch grammar analysis work is necessary in order to complete the documentation of Ega and to present it in an appropriate format for dissemination to scholars as planned. Intermediate results are being communicated to the community as they are produced.

As already noted, in parallel to this work, which is being performed at Bielefeld and Abidjan, the methodology in which local Abidjan personnel are being trained will be applied to two other languages in the Ivory Coast which fulfil the criteria of endangerment, scientific interest, and potential benefit to the community: Mbre, a newly discovered language of unknown affiliation in Central Ivory Coast, and the final remaining dialect of Eotile, a Kwa language whose other dialects are already extinct, but whose remaining dialect is used by a fishing community on Aby Lagoon.

Mbre

Mbre (Bre, Bere, Pré Pisia) is one of the most recently discovered language in the Niger-Congo area. Four years ago, the existence of the Mbre language was entirely unknown, and the language still does not figure among the languages reported by the recent edition of Ethnologue. The affiliation of Mbre to other languages in West Africa, the structure of the language, and the specific properties of the culture of its speakers, are not known.The language is currently thought to be spoken only in 2 villages, that is, Bondosso and Niantibo, rather than the 7 villages where the language was originally found to be spoken. There are thought to be about 200 speakers of the language in its fully fledged form, but around 700 people recognize their Mbre ethnicity. The language is located at 20 km from Marabadiassa, north-west of Bouake and Diabo (cf. the Figure with the hand-drawn map by Jacques Rongier).

A number of preliminary fieldwork expeditions have been undertaken by colleagues in Abidjan in cooperation with Prof Denis Creissels, Lyons, and including Prof. Jérémie Kouadio, who is also contributing to the Ega project, and Dr Jaques Rongier, an informal adviser to the project. However, the language has not been documented at all in the sense of the term used here, and the small amount of initial information in traditional format indicates that the language is very different from any of the surrounding Gur languages or the not-too-distant Kwa languages.

We have established good working relations with the Mbre community, and base our estimation of the scientific, historical and cultural value of the language on the data collected so far:

• Kouadio has recorded 400 sentences based on a syntax questionnaire, originally developed by Creissels, which has been used for Ega.

• Rongier has collected 2000 words with French gloss, including a number of preliminary tables of grammatical paradigms, and 7 transcriptions of texts, including 6 dialogs of 10 sentences each, translated from French into Mbre. No documentation of the type required in the present project exists as yet.

Eotile

The Eotile language, as it is generally known to Ivory Coast linguists, and also called Betibe, is listed in the Ethnologue catalogue as Beti:

BETI (EOTILE) [EOT] 3,181 (1988 census). Southern Department, villages of Vitre I and Vitre II, Subprefecture of Grand Bassam. Niger-Congo, Atlantic-Congo, Volta-Congo, Kwa, Nyo, Potou-Tano, Tano, Western. The last speaker of the 'pure' form of Beti died about 1993. Only a few old men rmemeber a few words from that dialect. Present speakers use a variety that is heavily influenced by surrounding languages. Speakers are bilingual in Anyin and Nzema. Christian, traditional religion, other. Survey needed.

The information is inexact; our own brief fieldwork indicates that this apparently applies to one dialect, the variant spoken in Adiake, which died with the last speakers in 1985 with the last speakers. Another variety of Betibe is spoken in Vitre I and Vitre II, and is highly threatened by the enclaving languages, Abure and Nzema. We have made sample digitally recorded documents from a family of speakers with full active knowledge of the language (Ahoua and Connell). Apart from our own recordings, and a small amount of information by Hérault in the Atlas des langues Kwa very little linguistic work has been done on Eotile; so far we are only aware of one small anthropological study (Perrot 1993).

Eotile oral history has it that the Eotile were colonised by the Anyi a century or so ago, who forbade the use of Eotile on pain of death, and that one group left the original langauge area near Grand Bassam and fled 200km eastwards. The Eotile themselves are a peaceful group and do not indulge in warlike or expansionist activities. The Eotile are a fishing community, unlike the Anyi, and it appears that in consequence of this the Eotile fishing vocabulary and culture has been preserved, even though the neighbouring Anyi and Abure have exercised a very strong influence on the language. Similar points apply to other aspects of Eotile culture, including traditional medicine, music, art and festivals.

G. Motivation for the Nigeria branch of the extension

Language endangerment in the Niger Delta

There are numerous endangered languages in Nigeria; Blench (in press) lists some 25 which he considers to be extremely endangered (see also Blench 1998), while Connell (1998a, in press) additionally identifies a number of languages which are seriously threatened in the Nigeria-Cameroon borderland. The Niger Delta, in Southeastern Nigeria, is one region where the threat to indigenous languages is as great as, if not greater than it is elsewhere in Nigeria (Williamson 1997).

The importance and dominance of the oil industry to the Nigerian economy has created an immense threat to the ecology and environment, culture and traditions of the Niger Delta. According to Prof Kay Williamson, all languages of the Niger Delta can be considered endangered in the short to medium term due to the influence of English and Nigerian Pidgin. Igbo as well has had a not inconsiderable influence in the region, though this influence may now be waning. All languages of the region (or any subset of them) are of linguistic interest for historical and language classification reasons; this is especially true of Defaka, and the Central Delta languages discussed below. The Ijoid group to which Defaka belongs is a distinct group within Niger-Congo and is enclaved by languages of the Benue-Congo family, a language family to which it is only distantly related. What little is known of it suggests Defaka to be the most distinct of Ijoid languages. With rapid changes affecting the regional ecology - both biological and human - the documentation of languages in this region must be seen as an urgent priority. Two languages, Defaka and Nkoroo, are selected for direct and attention in this aspect of the project. A number of other languages are identified and a strategy by which they too can be given urgent attention is presented.

Endangered languages of the Niger Delta region: (1) Defaka

Defaka is a highly endangered language spoken by fewer than 1,000 people, with most speakers being over 40. The current edition of Ethnologue describes Defaka as follows:

DEFAKA (AFAKANI) [AFN] 1,000 or fewer (1992 Crozier and Blench). Rivers State, Bonny LGA, in the Niger Delta, town of Nkoro. Niger-Congo, Atlantic-Congo, Ijoid, Defaka. Related to, but a separate language from the Ijo group. Survey needed.

Blench (in press) estimates the number of speakers to be 200, indicating a fairly rapid decline in the last decade. It is spoken in one ward of the Niger Delta town of Nkoroo (Bonny LGA Rivers State, southeast of Port Harcourt), and the small isolated settlement of Iwoma Nkoroo, near Kono, an Ogoni town. This settlement is enclaved by Kana (Ogoni) speakers. Defaka is related to, but a separate language from, the Ijo group (Ijoid, Niger-Congo). The Defaka speak Nkoroo as their principal language, but most if not all Defaka also have fluency in Nigerian Pidgin and English, as well as other neighbouring languages on an individual basis. Until relatively recently, Igbo exerted a strong influence in the region, and though it may still be used to some extent as a trade language, this function has largely been replaced by Pidgin, and Igbo is no longer used as a medium of instruction in the schools.

Defaka has been the subject of two preliminary studies, Jenewari (1983) and Shryock et al (1996/7). Jenewari's report, in particular, makes clear the linguistic interest and importance of Defaka. The language is at once a potential key to crucial aspects of the linguistic history of West Africa (indeed sub-Saharan Africa generally), but is also the possessor of a number of structural features that are exceedingly rare in African languages. On the basis of the brief comparative work undertaken by Jenewari, it has the Ijo language cluster as its closest relative, but this relationship itself appears rather distant, making Defaka coordinate with the Ijo cluster in an Ijoid grouping as a separate branch of Niger-Congo. Defaka shares no particular relationship with any of the other neighbouring languages, apart from Ijo, other than that they all belong to the Niger-Congo phylum. Among the striking structural characteristics of Defaka are its SOV (Subject-Object-Verb) word order, like Mande and Gur languages, and its sex-based system of grammatical gender. While both of these traits are shared with other Ijoid languages, they are extremely rare elsewhere in Niger-Congo, and the details of their functioning in Defaka - such as are known - are sufficiently different as to make it likely that these are inherited traits in Defaka and not a result of contact with other languages. The documentation of Defaka must therefore be considered an urgent priority, as it can tell us much about both the history of Niger-Congo, the worlds largest language phylum, and about a grammatical system which may turn out to be unique within this phylum.

Defaka has in all likelihood come under influence from Nkoroo, the dominant language among the Defaka. The extent of this influence is at present unknown, but oral traditions reported from both groups indicate that they have been in contact for a considerable period of time, likely several generations. It would not be possible to do a proper study of Defaka without also studying Nkoroo, in order to establish which characteristics of Defaka are truly its own, and which result from the influence of Nkoro on Defaka.

Endangered languages of the Niger Delta region: (2) Nkoroo

Almost paradoxically, the enclaving language Nkoroo is itself endangered in the short or medium term. Its study and documentation should not be postponed. The Ethnologue description is:

NKOROO (NKORO) [NKX] 4,550 (1989 UBS). Rivers State, Bonny LGA. Niger-Congo, Atlantic-Congo, Ijoid, Ijo, Eastern, Northeastern, Nkoroo. A separate language within the Ijo cluster. Survey needed.

The 1963 Nigerian census reported Nkoroo to have a population of 20,000. The current edition of the Ethnologue (Grimes 2000) reports it to have approximately 4,500 speakers, based on figures from 1989. Even if these figures were to be considered only estimates, they clearly indicate a serious and dramatic decline in the number of speakers of the language. Nkoroo is spoken principally in Nkoroo town and a few outlying hamlets. Like their Defaka neighbours, the Nkoroo also speak English and Pidgin and Igbo to varying degrees, as well as other nearby languages on an individual basis. There is no published work on Nkoro, though it figures as one of several Ijo languages in comparative work in progress by Williamson (Williamson, in prep). This work is sufficient to show Nkoroo is a distinct language within Ijo.

In summary, both Defaka and Nkoroo meet the endangerment criteria specified by the VWS-DOBES program. Both are spoken by small and diminishing numbers of speakers; Defaka is no longer spoken in public, and indeed can no longer be considered a principle home language as it is not passed on to the younger generations. Nkoroo is the principle home language of both ethnic groups. It is to some extent also spoken outside the home, but its functions in this respect are now severely diminished. For both the Defaka and the Nkoroo, English and Nigerian Pidgin are the primary public languages, a situation that has been imposed by economic and political pressures. There is an urgent need to document these languages, due to both their severely endangered status and their inherent linguistic interest. The time to do this work is now, while enough speakers with sufficient competence in the languages are available to still permit comprehensive documentation.

I. Summary of endangerment status

H. Official research permission and local community consent

Ivory Coast: research permission is granted by our partner institution, the Institut de Linguistique Appliquée at Université de Cocody, the Director of which, Dr François Kipré Blé is a former DAAD funded visiting scholar in Bielefeld and a strong supporter of the project. The local speech communities have been visited by several colleagues, and have shown themselves to be very interested and cooperative in preliminary fieldwork activities.

Nigeria: Permission is granted by the Nigerian Universities Commission and is automatic when collaborating with scholars at Nigerian institutions (in this case the Universities of Uyo, Port Harcourt, Calabar); the proposers have long-standing cooperation with Nigerian universities. Contacts by local scholars show that the local speech communities are very interested and cooperative.

J. Summary of documentation goals

In the pilot phase proposal, the computational documentation techniques to be employed were justified in detail. These techniques turned out to complement very well the methodology of the database project TIDEL. The techniques will be further developed and deployed, using the specific tools which were developed within the Ega project, with advice and feedback from partners in the DOBES consortium.

A. Main computational contributions to documentation:

1. Workbench for automatic concordance construction from annotated signal data.

2. Format conversion programs to enable local use of different, perhaps platform specific labelling data.

3. Automatic generation of documentation in XML format.

4. Automatic generation of hyperlexicon with multimodal information.

B. Main descriptive contributions to documentation:

1. Provision of multimodal digital signal data from

• speech, as audio data,

• speech, as laryngograph data,

• speech, as airflow data,

• gesture, posture, interaction (video).

1. Provision of word-level annotation of the signal data.

2. Provision of sociolinguistic questionnaire based interview data on language endangerment factors.

3. Provision of linguistic sketch descriptions (phonology, morphology, syntax).

4. Provision of a lexicon of significant size, including culture-specific terminologies.

Details of pilot phase work are summarised above, and documented in the project reports.

As in the pilot phase, the project will coordinate closely with the leading international initiatives associated with the international EAGLES/ISLE project, in which (including its predecessor projects) Universität Bielefeld has been an active player for over a decade.

For further details it will be convenient to refer to the proposal for the pilot phase project, as well as to the accompanying project reports.

1. Work schedule, time line and synchronisation

The tasks are essentially identical to the tasks which have been designed and tested during the pilot phase, with the difference that they have been upscaled in order to be able to test their suitability for producing more documentation of the required quality within he allotted time span.

Particulaly detailed attention has been paid to the coordination procedures for ensuring that ongoing work is optimally synchronised. These procedures were developed and applied during the pilot phase, and are based on over a decade of experience with procedures for organisation and coordination in large international standardisation and language/speech technology consortia (EAGLES, VerbMobil).

One of the main coordination strategies is telecooperation via email and the Bielefeld internet site, which also hosts the web site of the West African Linguistics Society. For this reason, adequate funding is requested for equipping the partners in Abidjan and Uyo with improved infrastructure for reliable and efficient email and internet download facilities (expenses in this area are orders of magnitude more expensive in West Africa than in Europe).

One organisational contstraint is due to climatic conditions: during the tropical wet season, most of the roads and paths in the regions concerned are flooded or swampy and impassable for expedition vehicles. Economical procedures dictate that fieldwork be conducted during the dry season. The structure of the university year in the countries concerned puts a further constraint on these activities; fieldwork is therefore in general limited to February, March and April. For local training sessions, August and September are also feasible.

A number of tasks are designed for joint work:

• cross-checking and joint evaluation of documentation results,

• joint training workshops,

• joint report authoring,

• joint publications, also with other partners in the funding programme.

In the first few months of the project, a detailed workflow design will be produced in cooperation with all partners, as in the pilot phase.

The following work plan is divided into three parts, Tasks, Task assignment and Time line, each of which is presented in tabular overview form.

A. Tasks

B. Task assignments to main project participants

C. Time schedule

Note: the time blocks do not represent effort, which is divided task dependently.

1. Calculation and substantiation of costs

removed

removed

removed

1. Curriculum vitae and publications of main participants

Prof. Dafydd Gibbon

Dafydd Gibbon is the initiator and coordinator of the present proposal. His qualifications are in general and computational linguistics, with special emphasis on the computational modelling of tone languages, and on applications of linguistics to the human language technologies. He has directed doctoral theses on languages on West African language families (Gur, Kwa) and has visited the University of Cocody, Abidjan, in 1986, 1991, 1996 and 1998 as DAAD guest lecturer and as guest of the University. He is an experienced field linguist, having worked on Baule (a major Western Kwa language), and Go (a little known South Mandé language, related to Dan/Yacouba). Prof. Gibbon has developed influential computational models of register tone systems (Ewe/Gbe, Baule/Kwa, Tem/Gur), and is currently developing a speech synthesis system for Anyi, a Kwa language closely related to Baule, as a source of operational criteria for the correctness of phonetic descriptions. He is a member of the Gesellschaft für bedrohte Sprachen. With Prof. Christian Lehmann (now U Erfurt) he co-proposed the DAAD funded joint project with U Cocody, Abidjan, Encyclopédie des Langues et Cultures de Côte d'Ivoire, in 1997, and has directed the project since April 1999. The project has provided basic training for local Ivorian personnel who will support the work described in the present proposal. Prof. Gibbon has had many years of experience in directing project consortia, including the lexicon subconsortium of the BMBF funded Verbmobil speech to speech translation project (1991 to present), the EAGLES (Expert Advisory Groups for Language Engineering Systems) Spoken Language Working Group Phase I (1994-1996) and Phase II (1996-1998). Among his publications, the major recent contribution to language and speech documentation issues is the Handbook of Standards and Resources for Spoken Language Systems (1997), Berlin: Mouton de Gruyter, of which he was the chief editor (with Roger Moore and Richard Winski). The Handbook has become the international standard reference work for technologically supported corpus collation and documentation, and system development and evaluation. The results of Phase II of the EAGLES project appeared in 2000 as the Handbook of Multimodal and Spoken Dialogue Systems (Kluwer Academic Publishers). Prof. Gibbon is a fluent speaker of French, including the Ivorian variety, the contact language for the Ega speech community.

Dr Bruce Connell

Bruce Connell is a linguist and phonetician with many years of experience in fieldwork and both traditional and multimedia language documentation, and an internationally renowned expert on endangered languages. In his thesis (U Edinburgh) Phonetic Aspects of the Lower Cross Languages and their Implications for Sound Change Dr Connell developed a new methodology in which fieldwork is coupled with experimental phonetic methodology, enabling questions of phonological change to be addressed in more detail than previously. Dr Connell has published extensively using this methodology and on problems of documenting endangered languages. Currently, Dr Connell is currently Visiting Professor of Linguistics at York University, Toronto and a Research Associate of the School of Oriental and African Studies, Unversity of London. He is a former ESRC Research Fellow at the Institute of Social and Cultural Anthropology, U Oxford, UK, and has been Director of the Phonetics Laboratory, U Oxford, Visiting Lecturer at U Newfoundland, Lecturer at U Calabar, Nigeria, and Graduate Assistant and Instructor at U Alberta, Edmonton. He has acquired and worked on the projects Language Contraction and Death in the Nigeria-Cameroon Borderland (ESRC), A Comparative Survey of Mambila Dialects (ESRC), was Vans Dunlop Scholar at U Edinburgh, and held a Fellowship for Second Language Study at U Ottawa. Dr Connell has presented his work at more than forty conferences in linguistics, phonetics, laboratory phonology, speech technology, African linguistics, and anthropology, is a member, of the International Phonetic Association, and a founder member of the Foundation for Endangered Languages. Dr Connell is a well-known and respected figure in West African linguistics, and is on the Council of the West African Linguistics Society. He is a fluent speaker of French, the relevant contact language for work in Ivory Coast. He has spent much of the past 18 years working in southeastern Nigeria and the adjacent areas of Cameroon, where he has conducted extensive fieldwork, much of which has focussed on endangered languages. He is the leading authority on the comparative linguistics of the Lower Cross group, one member of which (Obolo) neighbours Defaka and Nkoroo.

Dr Firmin Ahoua

Dr. Firmin Ahoua is a member of the Institut de Linguistique Appliquée and Associate Professor at the Département de Linguistique, both at the Université de Cocody, Abidjan, Côte d'Ivoire. Dr Ahoua was educated at the Department of German of that university (then Université Nationale de Côte d'Ivoire) and the Universities of Saarbrücken (M.A.) and Bielefeld (Ph.D.) as DAAD scholar. Subsequently he held a Fulbright Scholarship at the University of California at Berkeley, and is currently Humboldt Scholar jointly at the Universities of Bielefeld and Stuttgart. He was co-director of the NSF funded project Comparative Phonology of Ivoirian Languages, concentrating on the Kwa/Tano group, in which extensive audio recordings of different primary data types, transcriptions and comparative phonological analyses were made, and doctoral dissertations supervised. Dr Ahoua has been Visiting Professor at the University of Ouagadougou, Burkina Faso, has held numerous positions in West African academic review and selection committees, and is an active member of the West African Linguistic Society. He has published numerous papers on various aspects of the phonology and discourse prosody of Kwa languages, and was local co-director of the Bielefeld-Abidjan Encyclopédie project. In Summer 1999, together with Dr Connell, Dr Ahoua did preparatory fieldwork on Ega. Dr Ahoua recently supervised an M.A. thesis on elision and vowel harmony in Ega (Dago 1999). Dr Ahoua is a native speaker of two Kwa languages, his maternal language Baule and his paternal language Attié, and French, and speaks near-native German and English.

Dr Eno-Abasi Urua

Dr Eno-Abasi Essien Urua is Senior Lecturer in the Department of Linguistics and Nigerian Languages, University of Uyo, PMB 1017, Uyo, Akwa Ibm State, Nigeria. Dr Urua received her training in linguistics at University of Calabar, University of Ibadan, University College London, University of Edinburgh, and has received numerous awards, including the Post-Doctoral Fellow Award of the University of Edinburgh and an Alexander-von-Humboldt Fellowship at Universität Bielefeld, as well as research grants from University College London and the University of Uyo. Dr Urua has numerous publications in a variety of linguistic and applied linguistic journals, including co-authored publications with a number of other internationally known scholars, including A. Akinlabi, Ruttgers University, and Steven Bird (University of Pennsylvania). Dr Urua's research interests cover all aspects of fieldwork oriented empirical linguistics in relation to the languages of Nigeria, focussing in particular on questions of phonology and prosodic structure.

From April 2000 to July 2001 Dr Urua was Humbolt scholar, hosted by Prof. Dafydd Gibbon at Universität Bielefeld, where she worked on linguistic problems of Southern Nigerian languages, in particular Lower Cross languages, focussing on Ibibio. During this time, she had a number of papers accepted at international conferences (e.g. Freiburg, Toulouse), some jointly with colleagues at Universität Bielefeld, and organised the widely acclaimed TAPS 2001 Workshop (Typology of African Prosodic Systems) at Universität Bielefeld, supported by Dr Ulrike Gut and Prof. Dafydd Gibbon of Universität Bielefeld. Dr Urua has also worked on comparative phonetic studies involving Ega and Anyi (Ivory Coast) and Ibibio (Nigeria).

1. Related project applications to other funding institutions

No related project applications have been made to other funding institutions.

A DFG funded project is expected to start in October 2001 in the framework of the DFG-Forschergruppe: Texttechnologische Informationsmodellierung (Teilprojekt: B-3, Theorie und Design multimodaler Lexika. This project is a basic research project, with no direct connection to the applied field of language documentation. However, it is expected that the present project will benefit from the general infrastructural environment provided by the Forschergruppe.

1. Planned symposia

Annual workshops/symposia are planned, and will be applied for separately. The first will be in cooperation with other funding programme partners, coordinated by the TIDEL project, in May 2002.

The Symposion already planned in the pilot phase under the aegis of the West African Linguistic Society took place informally in summer 2000. An application for funding in the years 2002 and 2004 (EUR 10000 for each meeting) is hereby applied for.

1. Other cooperating institutions

The cooperations detailed in the pilot phase proposal have been pursued as planned. Particularly close cooperation has been developed with Prof. Will Leben, Stanford University, and Dr Steven Bird, University of Pennsylvania and Linguistic Data Consortium.

1. Archive of relevant publications

The following literature list contains selected relevant publications by the project scientists, and on the languages to be documented.

Publications Ahoua

Ahoua, F. (1986). The autosegmental representation of tones in Akan: evidence for the tone mapping rules. In: K. Bogers, H. v.d.Hulst & M. Mous, ed., The Phonological Representation of Suprasegmentals. Dordrecht: Foris.

Ahoua, F. (1990). Two current phonetic models in intonation analysis. Cahiers Ivoiriens de Recherche Linguistique 25.

Ahoua, F. (1996). Prosodic aspects of Baule. Köln: Rüdiger Köppe Verlag.

Ahoua, F. (2000). Reconstruction of the Prosodic System of Kwa Languages of Ivory Coast. Habilitation monograph (submitted), Universität Bielefeld.

Ahoua, F. & E. Gbéry (1991). Aspects of the tonal morphophonology of Abbey. Cahiers Ivoiriens de Recherche Linguistique 26.

Ahoua F. & W. Leben (1994). High tone sequencing in Baule. In: F. Ingemann, ed. Proceedings of the Mid-America Linguistics Conference. Lawrence, Kansas: MALC Publications.

Publications Connell

Connell, B. (1987). Language and pre-history in the Cross River basin. In M.B. Abasiattai (Ed.) The Role of the Arts in Nation Building. University of Calabar Faculty of Arts Occasional Publications, No. 1. Calabar: Map Publishing, pp. 49-64

Connell, B. and K.B. Maison (1994). A Cameroun homeland of the Lower Cross languages? SUGIA (Sprache und Geschichte in Afrika) Vol. 15: 47-90.

Connell, B. (1994). The Lower Cross Languages: a prolegomena to the classification of the Cross River languages. Journal of West African Languages, 24 1: 3-46

Connell, B. (1994). The language situation in Nigeria. In R.E. Asher (Gen. Ed.), Encyclopedia of Languages and Linguistics. Oxford: Pergamon.

Connell B. (1998). Moribund languages of the Nigeria-Cameroon borderland. Brenzinger, M. (Ed.), Endangered Languages in Africa. Cologne: Rüdiger Köppe. Pp. 207-225.

Connell, B. (1998). Yams, palms, and culture history in the Cross River Basin: Linguistic evidence. R. Blench and M. Spriggs (Eds.), Archaeology and Language, Vol. II. Routledge One World Archaeology Series. London: Routledge. Pp. 324-365.

Connell, B. (1998). Classifying Cross River. Maddieson, Ian & Hinnebusch, Thomas J. (Eds.), Language History and Linguistic Description in Africa: Trends in African Linguistics Vol. 2. Lawrenceville, NJ: Africa World Press. Pp. 17-25.

Connell, B. 1998. Mambilex: A lexical database of the Mambiloid languages. Lexical database of some 30 languages and dialects with audio, phonetic and phonological transcriptions, morphological information.

Connell, B. (2000) Factors in language attrition in Africa. Paper presented to Against All Odds: African Languages and Literatures in the 21st Century Asmara, Eritrea.

Connell, B. (in press) Language endangerment in Central Africa. In M. Brenzinger (Ed.), Language Diversity Endangered Berlin: Mouton de Gruyter.

Connell, B. (in press) Lexicography, Linguistics and Minority Languages. To appear in JASO (Journal of the Anthropological Society of Oxford).

Connell, B. (to appear) Phonetic/Phonological Variation and Language Contraction. International Journal of Sociolinguistics.

Connell, B., Ahoua, F., & Gibbon, D. (to appear) Ega: A preliminary assessment of endangerment. In L. M. Hyman, & I. Maddieson (Ed.), Trends in African Linguistics Vol. 7. Trenton NJ: Africa World Press.

Connell, B., Ahoua, F., & Gibbon, D. (submitted). Ega. Journal of the international Phonetic Association.

Connell, B., Ahoua, F., & Gibbon, D. (in preparation). Sociolinguistic tools for the assessment of language endangerment.DOBES Technical Report X (Ega).

Publications Gibbon & Bielefeld team

Gibbon, Dafydd (1999a). Computational lexicography. F. Van Eynde & D. Gibbon, eds., Lexicon Development for Speech and Language Processing. Dordrecht: Kluwer Academic Publishers.

Gibbon, Dafydd (1999b). EAGLES Final Report. University of Bielefeld. Final Report of LE EAGLES Phase II project (LE3-4244 10484/0) for the European Commission.

Gibbon, Dafydd (2001). On lexical objects and their properties. a contribution to the `metalex' requirements specification for spoken language lexicon documentation. Universität Bielefeld: DOBES Technical Report (Ega).

Gibbon, Dafydd (2001, in press). Finite state prosodic analysis of african corpus resources. Proceedings of Eurospeech 2001, 1:79-82, Aalborg. EUROSPEECH.

Gibbon, Dafydd & Ulrike Gut (2001, in press). Measuring speech rhythm. Proceedings of Eurospeech 2001, 1:91-94, Aalborg. EUROSPEECH.

Gibbon, Dafydd, Silke Kölsch, Inge Mertins, Michaela Schulte & Thorsten Trippel (1999). Terminology principles and support for spoken language system development. Proceedings of EUROSPEECH '99, Budapest. EUROSPEECH.

Gibbon, Dafydd, Harald Lüngen, Andreas Witt (2000a). Enhancing speech corpus resources with multiple lexical tag layers. Proceedings of LREC 2000, Athens.

Gibbon, Dafydd, Inge Mertins & Roger Moore (2000b). Handbook of Multimodal and Spoken Dialogue Systems: Resources, Terminology, and Product Evaluation. Dordrecht/New York: Kluwer Academic Publishers.

Gibbon, Dafydd, Roger Moore & Roger Winski, (1997). Handbook of Standards and Resources for Spoken Language Systems. Berlin: Mouton de Gruyter.

Gibbon, Dafydd & Thorsten Trippel (2000). A multi-view hyperlexicon resource for speech and language systemdevelopment. Proceedings of LREC 2000, Athens.

Gibbon, Dafydd, Thorsten Trippel & Serge Sharoff (2001, in press). Concordancing for parallel spoken language corpora. Proceedings of Eurospeech 2001, 3, 2059 - 2062, Aalborg.

Trippel, Thorsten, Nils Jahn & Soma Ouattara (2001a). Preliminary Specification, Design and Proof-of-Concept Implementation of a Portable Audio Concordance (PAC). Universität Bielefeld.

Trippel, Thorsten, Nils Jahn, Thurid Spiess, Dafydd Gibbon & Soma Ouattara (2001b). Portable Audio Concordance System (PAX)Integration of Acquisition, Search and Enhanced Display Modules. Universität Bielefeld.

Trippel, Thorsten, Soma Ouattara & Nils Jahn (2001c). Proof-of-concept implementation of a portable audio concordance. http://www.spectrum.uni-bielefeld.de/langdoc/.

Van Eynde, Frank & D. Gibbon, (1999). Lexicon Development for Speech and Language Processing. Kluwer, Dordrecht.

Publications Urua

Akinlabi, A. & E. E. Urua 2000. Tone in Ibibio verbal reduplication. Proceedings of the 2^nd World Conference of African Linguistics Leipzig 1997 (eds.) H. Ekkehard Wolff & Orin D. Gensler, 279-291.

Urua, E. E. 1987. Segment deletion and aspects of tone in Ibibio. MA project. University of Ibadan, Ibadan, Nigeria.

Urua, E. E. 1990. Aspects of Ibibio phonology and morphology. Ph.D Dissertation. University of Ibadan, Ibadan, Nigeria.

Urua, E. E. 1995. The status of contour tones in Ibibio. Theoretical approaches to African Linguistics (ed.) Akinbiyi Akinlabi, 329-343.

Urua, E. E. 1996. Languages or dialects? The linguistic situation in Akwa Ibom State. Journal of Humanities vol. 4, 50-65 (ed.) David Eka.

Urua, E. E. 1996/1997. A phonetic analysis of Ibibio tones. Journal of West African Languages vol. XXVI (1), 15-26.

Urua, E. E. 1997. Object position in Eastern Lower Cross. Object position in Benue-Kwa. Victor Manfredi and Rose-Marie Dechaine (eds.), HIL Publications, The Hague: Holland Academic Graphics, 189-206.

Urua, E. E. 2000. Ibibio phonetics and phonology. Cape Town, South Africa: Centre for Advanced Studies of African Society.

Urua, E. E. and S. Bird 1994. The phonetic description of Ibibio tones. Proceedings of the University of Edinburgh Dept. of Linguistics Conference, 210-216.

Gibbon, Dafydd, Eno-Abasi Urua & Ulrike Gut (2000). How low is floating low tone in Ibibio? Presented at the 30^th Conference on African Linguistics. Leiden.

Urua, E.E. 2000). The rhythm of tone languages: a study of Ibibio, Ega and Anyi. Presented at 1^st international workshop on Typology of African Prosodic Systems (TAPS), Universität Bielefeld.

Bibliography Ega

See Appendix, Catalogue of documents

Bibliography Eotile (Beti, Betibe), Mbre (Bre, Bere, Prepisia), Ivory Coast

Dumestre, G. (1970). L'éotilé. In: Atlas Linguistique de Côte d'Ivoire, Les Langues de la Région Lagunaire. ILA, Universite d'Abidjan.

Goprou, D. Carlos. to appear. Morphophonologie de l'éotilé. in: Atlas phonologique des langues Kwa de Côte d´Ivoire ed. by F. Ahoua and W. Leben.

Paris: Documents de Linguistique Africaine.

Herault, G. (1982). L'Eotilé. In: G. Hérault, ed., Atlas des langues Kwa de Cote d'Ivoire. Abidjan: Institut de Linguistique Appliquee, ACCT. 404-424.

Retord, G (1970). Le domaine linguistique éotilé. Bulletin de Liaison du CURD, 1.

Perrot C. -H., 1993, ‘Prophétisme et modernité en Côte d’lvoire. Un village éotile et le culte de Gbahie’, in J. -F. Bayart (ed.), op. cit., pp. 215-275.

Rongier, Jacques (1997). Histoire de Niantibo. Ms. (Mbre)

Rongier, Jacques (2001). Enquêtes à Niantibo, Bondosso et Bere-Ko. Ms. (Mbre)

Stewart, J. (1971). Niger-Congo: Kwa. In: T.A. Sebeok, ed., Current Trends in Linguistics 7: Linguistics in Sub-Saharan Africa. Den Haag: Mouton.

Bibliography Defaka, & Nkoroo (Ijoid, Cross-River, Nigeria)

Blench, R. M. (1998) The status of the languages of Central Nigeria. In M. Brenzinger (Ed.), Endangered Languages in Africa (pp. 187-205). Köln: Rüdiger Köppe Verlag.

Blench, R. M. (in press) Language death in West Africa. In M. Brenzinger (Ed.), Language Diversity Endangered Berlin: Mouton de Gruyter.

Connell, B. (1994) The Lower Cross languages: a prolegomena to the classification of the Cross River languages. Journal of West African Languages, XXIV 1, 3-46.

Connell, B. (1995) The historical development of Lower Cross consonants. Journal of African Languages and Linguistics, 16 1, 41-70.

Connell, B. (1998a) Moribund languages of the Nigeria-Cameroon borderland. In M. Brenzinger (Ed.), Endangered Languages in Africa (pp. 207-225). Köln: Rüdiger Köppe Verlag.

Connell, B. (1998b) Classifying Cross River. In I. Maddieson, & T. J. Hinnebusch (Ed.), Language History and Linguistic Description in Africa. Trends in African Linguistics Vol. 2 (pp. 17-25). Trenton NJ: Africa World Press.

Connell, B. (in press) Language endangerment in Central Africa. In M. Brenzinger (Ed.), Language Diversity Endangered Berlin: Mouton de Gruyter.

Gardiner, I. (1980) Abuan-English, English-Abuan Dictionary. Port Harcourt and Jos: University of Port Harcourt Press and the Nigerian Bible Translation Trust.

Grimes, B. F. (2000) Ethnologue (14th ed.). Dallas: S.I.L.

Jenewari, C.E.W. (1983) Defaka: Ijo's closest linguistic relative. Port Harcourt: University of Port Harcourt Press. (Also in I.R. Dihoff (Ed.) Current Approaches to African Linguistics Vol 1 pp 85-111 in an abridged form.)

Jenewari, C.E.W. (1989) Ijoid. In J. Bendor-Samuel (Ed.), The Niger-Congo Languages (pp. 105-118). Lanham: University Press of America.

Nakagawa, H. (1992) A Ban vocabulary. Journal of African and Asian Studies, 4 3, 113-148.

Shryock, A., Ladefoged, P., & Williamson, K. (1996/97) The phonetic structures of Defaka. Journal of West African Languages, XXVI 2, 3-27.

Williamson, K. (1985) How to become a Kwa language. In A. A. M. Makkai, Alan K. (Ed.), Essays in Honor of Rulon S. Wells (pp. 427-443). Amsterdam: John Benjamins.

Williamson, K. (1997) Endangered languages of the Port Harcourt region of Nigeria. Paper presented to the Symposium on Language Endangerment on Africa, Leipzig, Aug 1997.

Williamson, K. (in preparation) Comparative Ijoid.

Wolff, H. (1969) A Comparative Vocabulary of Abuan Dialects. Evanston: Northwestern University Press.

1. APPENDIX: Extract from provisional documentation database