Working in this context is very interesting because all the necessary resources exist at the European Parliament : members speeches in their own language, their (speech) interpretation in the different languages of the Parliament, their transcription into written form, and the translation of the transcripts in the different official languages. Thus, these data allow for training the automatic interpretation systems, including recognition in the source language, translation from the source language to the target language, and speech synthesis in the target language, thus utilising both monolingual and crosslingual technologies. A demonstration of a system for the English-Spanish language pair is available on line 20. TC-Star has also produced and distributed a report in five languages on Language Technology in Europe [LAZZARI, STEINBISS, 2006] 21.
15 http://ec.europa.eu/commission_2004-2009/orban/index_en.htm 16 http://ec.europa.eu/education/policies/lang/doc/multireport_en.pdf 17 http://europa.eu/rapid/pressReleasesAction.do reference=IP/08/1340&format=HT ML&aged=0&language=EN&guiLanguage=en 18 http://www.europarl.europa.eu/sides/getDoc.dotype=TA&language=EN&referenc e=P6-TA-2009-19 http://www.tc-star.org 20 See “Demo JM.asf” on the website http://audiosurf.org/demo_video 21 http://www.tc-star.org/pubblicazioni/D17_HLT_ENG.pdf Joseph Mariani In the seventh European Framework Programme, fp7 (2007-2013), this area is mainly conducted by the “Language Technology, Machine Translation” Unit. In addition to R&D projects, an infrastructure and two networks have been established : clarin (Common Language Resources and Technology Infrastructure) 22, FLaReNet (Fostering Language Resources Network) 23, and meta-net (Multilingual Europe Technology Alliance) 24.
clarin is an infrastructure supported by the programme esfri (European Strategy Forum on Research Infrastructure) of the European Commission.
Its objective is the distribution of language resources and tools for Humanities and Social Sciences.
FLaReNet is a Thematic Network supported under the e-Content European Programme, with a budget of ˆ0.9 million over 3 years (20082011). Its purpose is to serve as a think tank for the promotion of language resources in European programmes.
The meta-net Network of Excellence was established within the t4me (Technologies for a Multilingual Europe) project. This project has a budget of ˆ6 million over a period of 3 years (2010-2013) and is structured in three parts :
– pushing the research frontiers in machine translation ;
– establishing an Open Resources Infrastructure (meta-share), including the production, annotation, standardisation, validation and distribution of language resources, and the evaluation of language technologies ;
– to conduct a reflection on the place of multilingual technologies in the context of drafting a Strategic Research Agenda for the next Framework Program (2014-2020).
EUROPEAN AND INTERNATIONAL PERSPECTIVE The resolutions of the European authorities demand a major effort to process all European languages, national and regional. However, if one considers the number of languages or language pairs that are to be 22 http://www.clarin.eu 23 http://www.flarenet.eu 24 http://www.meta-net.eu Joseph Mariani addressed, and multiply it by the number of technologies, we see that the size of the effort is probably too large for the European Commission alone. It would therefore be interesting to share this effort among Member States, or Regions, and the European Commission, in perfect harmony with the “principle of subsidiarity”.
Language technologies are well suited for a joint effort. The European Commission would have the primary responsibility for overseeing and ensuring coordination of the programme (management, provision of standards, technology evaluation, communication…) and of developing core technologies around language processing. Each Member State would have as a priority to ensure the coverage of its language(s) : to produce the language resources essential for the development of systems (corpora, lexicons, dictionaries), and to develop or adapt technologies to the specificities of its language(s). This model would be easily adaptable to an international effort, combining the efforts of the participating countries and of international organizations.
Unfortunately, until now the topic of Language Technologies has been regrettably considered just as one research area among many others in Europe, not as an essential element of the the European construction, requiring a high priority effort to handle the corresponding issues. This weakness is all the more dangerous given the liveliness of the European Union and its needs to increase economic, informational and cultural exchanges between countries, and to address the citizens of each State and help them in their communication. Let’s hope that the political awareness of the issues attached to multilingualism will see research in language technologies receive adequate attention in future Framework Programmes.
CONCLUSION Language technologies are a major tool to facilitate multilingualism in Europe as well as in the rest of the world. To achieve this, we need to agree to coordinate the efforts of States, even regions, and international organizations (European Commission, United Nations, Unesco, the African Union, etc.), involving industry and public research laboratories. Care should be taken to produce for each language the language resources Joseph Mariani needed, and organise the research effort in an open way, based on the interoperability and objective benchmarking of technologies.
We could then add a nod to the famous phrase of Umberto Eco, saying :
“Translation is the language of Europe… with the support of technology”, and extend this assumption to the global village.
BIBLIOGRAPHY [CENCIONI, ROSSI 2008] R. Cencioni, K. Rossi. Language based Interaction, EC-ICT Conference, Lyon, 26 Novembre 2008.
[CHAUDIRON, MARIANI 2006] S. Chaudiron, J. Mariani. Techno-langue: The French National Initiative for Human Language Technologies (HLT), Proceedings LREC’06, Genoa, Italy, May 2006.
[ECO 1993] U. Eco. La langue de l’Europe, c’est la traduction, Assises de la traduction littraire, Arles, 1993.
[GASQUET-CYRUS, PETITJEAN EDS 2006] M. Gasquet-Cyrus, C. Petitjean eds. Le poids des langues, L’Harmattan, 2009.
[KOEHN, BIRCH, STEINBERGER 2009] Ph. Koehn, A. Birch and R. Steinberger. 462 Machine Translation Systems for Europe, Machine Translation Summit XII, p. 65-72, 2009.
[LAZZARI, STEINBISS 2006] G. Lazzari, V. Steinbiss. Human Language Technologies for Europe, TC-Star Report, April 2006.
[MARIANI 1995] J. Mariani, ed. Evaluation chapter in Survey of the State of the Art in Human Language Technology, R. A. Cole, J. Mariani, H. Uszkoreit, N. Varile, A. Zaenen, A. Zampolli, V. Zue eds., Cambridge University Press, 1995.
[PAPINENI, ROUKOS, WARD, ZHU] K. Papineni, S. Roukos, T. Ward, W.-J. Zhu. BLEU: A Method for Automatic Evaluation of Machine Translation. In : Proceedings of the 40th Annual Meeting of ACL, Philadelphia, PA.
Joseph Mariani VASSILI RIVRON THE USE OF FACEBOOK BY THE ETON OF CAMEROON Opened in 2006, Facebook currently has hundreds of millions of users worldwide, who use several dozen languages. Taking the example of Facebook groups run by and for members of the ethnic group ton (Central Cameroon, approximately 250 000 speakers), we can observe how the use of the latest communication technologies enables new ways to affirm traditional culture.
Original article in French.
Translated by Laura Kraftowitz.
VASSILI RIVRON is an anthropologist, lecturer at the University of Caen Basse-Normandie. He is attached to CERReV (UCBN) and associate researcher at the cole des Hautes tudes en Sciences Sociales and at CSU (CNRS) acebook, a “social network” launched in 2006, has gained hundreds of millions of users worldwide. The site interface is available in Fdozens of languages (including regional languages like Basque and artificial languages like Esperanto), and uses multiple writing systems, alphabetical or other. It allows for diverse uses, among which one finds projections onto the network web of pre-existing cultural practices and references, but also of cultural innovation.
By considering the example of several “Facebook groups” that are run by and for members of the ethnic group the Eton (located in Central Cameroon, counting approximately 250,000 speakers), we can observe how the use of recent communication technologies has enabled new forms of affirmation of traditional cultures. One observation is the extension of a mother tongue outside its habitual context and uses, and the correlated development of its graphic system.
In a country with two official languages (English and French) and over two hundred local languages, the Eton 1 language appeared as though it would remain, as many others, confined to its specific geographical and social context. These specific contexts are the inhabitants of the Lki department, and domestic space, family circles, and traditional hierarchies for the urban migrants and emigrants ; and all this principally within the context of a interpersonal communication (including the telephone), or at least in the presence of interlocutors (i.e. speeches).
Some Cameroonian languages are or have been written (including by specific graphic systems). Some were taught during the period of the German protectorate (1884-1922), before this was abolished under French colonization. Proposals to teach the local languages in public schools have 1 A tonal Bantu language, located to the North of the Beti/Bulu/Fang linguistic complex.
Vassili Rivron periodically made their way onto the national agenda since independence (1960), but have been blocked by a republican imaginary that fears tribal schisms. Eton thus continues to enjoy a primarily oral existence, its graphic system remaining uncodified. Its alphabet appears mainly in scholarly works : ethnographic transcriptions, linguistic studies, bilingual folklore collections and oral literature compilations. It also serves as a mnemonic resource to its users for personal annotations (journals, sayings, notes) or in passages of collusion or scholarship embedded within correspondence that is written principally in another language 2. The lack of education and the republican principles inherited from France explain why regular readership in Eton is impossible (in print media, books, government, and politics).
From the very beginning of our study (2004), we were able to observe an extension of Eton written on the web, mainly in two contexts. One was the creation of a number of cultural heritage sites (on folklore, culture, language, ethnic group or regional history), which were often individual initiatives. On these sites, Eton might appear in fragmentary lexicons, or collections of sayings, for example. These sites rarely addressed themselves directly to the public in written Eton, but rather transcribed or taught the language’s oral form. For obvious reasons, written dialogue in Eton appeared mainly in blogs and forums, and were especially prominent in the comment sections of YouTube videos by local artists and Cameroonian online news sites. But in both cases, the writing rarely exceeded two sentences without being translated and often boils down to illustrations, posing questions, witticisms or talk of complicity (hence excluding non-speakers).
The emergence of social networks (i.e. Facebook) has prompted the development of a wide variety of communities (“groups”), which generally overlap individual entities (photo and personal data) with cooptation procedures (“friend requests”), and which provide access to “profile” or “group” contents (photos, texts, videos, games, etc.) that can be shared and customized. For our purposes, we highlight several interesting Facebook groups : The Etons (ethnic referent, 53 members), Sons and Daughters of Lki (territorial referent, 220 members), and Ongola – Fang-Bulu-Beti 2 For more on techniques and languages of correspondence with the diaspora, see Sayad, Abdelmalek, “Du message oral au message sur cassette : la communication avec l’absent”, Actes de la recherche en sciences sociales, n°59, 1985, pp. 61-72.
Vassili Rivron Culture (enlarged cultural referent, 1,445 members). Apart from transposing various aspects of the groups’ social structure and cultural practice onto a digital resource, we also find forms of writing that are evolving in comparison to our initial observations. Among the many topics covered in these Facebook groups, a significant portion is devoted to generic issues relating to “traditional” culture, marriage, parenthood, initiations and sayings. The question of language is rarely specifically addressed.
Sustainable exchanges are established in the social networks, using different registers. French and English are clearly dominant for generic substance (group description and instructions), or for comments and dialogues themselves. But at the heart of personal messages, profile “walls”, and discussion forums, Eton may be used for section titles, opening and continuing a debate (other languages may also be included).
There seems to be less hesitation to write in Eton in the “among friends” atmosphere of Facebook groups, where mutual understanding is postulated despite the absence of codification and official teaching of customary spelling. The inequality of language skills is evident in these exchanges, along with the diversity of resources mobilized to design graphic solutions (in Francophone, Anglophone, and Ewondophone environments 3).
Efforts to make themselves understood in writing in this tonal language, for example to decrypt messages by sounding them out, are visible in exchanges and during our direct observations of users. The pleasure of this activity is also clear ; it is an expression of solidarity and cultural pride.
Of interest to the researcher is that this textual corpus is supplemented by “profiles” that provide valuable information on active group members.