Researching Less-Resourced Languages – the DigiSami Corpus

Researching Less-Resourced Languages – the DigiSami Corpus Kristiina Jokinen University of Helsinki, Finland and AIRC, AIST Tokyo Waterfront, Japan kristiina.jokinen@helsinki.fi Abstract Increased use of digital devices and data repositories has enabled a digital revolution in data collection and language research, and has also led to important activities supporting speech and language technology research for less-resourced languages. This paper describes the DigiSami project and its research results, focussing on spoken corpus collection and speech technology for the Fenno-Ugric language North Sami. The paper also discusses multifaceted questions on ethics and privacy related to data collection for less-resourced languages and indigenous communities. Keywords: corpus collection, under-resourced languages, North Sami 1. Introduction Several projects and events have increased research activities for under-resourced languages during the past years. For instance, the DLDP-project (Digital Language Diversity Project) is to advance the sustainability of Europe’s regional and minority languages, while the Flare- net network and the LRE Map (Calzolari et al. 2012) have had a big impact on sharing language resources and making speech corpora freely available. However, there are many challenges that researchers and developers face when aiming at the same technology standards for less-resourced and endangered languages as those for majority languages. The challenges do not only concern scarce data and non- optimal algorithms, but also issues inherently related to the cultural contexts of linguistic communities in general, and shared background of minority cultures in particular. It is thus important to pay attention to community-based techniques in data collection and technology development (crowdsourcing in a wide sense) as well as trying to connect the communities with other small language communities (Soria et al. 2013). An important aspect of such work is to empower minority language speakers with the knowledge and skills to create and share content for digital devices using their own language. The DigiSami project (http://www.helsinki.fi/digisami/) is supported by the Academy of Finland in a wider context of a joint initiative with the Hungarian Academy of Sciences. The focus of the Digital Citizens framework project was to support collaborative research on endangered Fenno-Ugric Languages and to develop tools and resources for automatic language processing as well as to experiment with new technology applications. The main motivation was to improve digital visibility and viability of the target languages, and to explore different choices for encouraging and maintaining the use of less-resourced languages in the digitalized world. The goals of the DigiSami project are discussed in Jokinen (2014) and Jokinen et al. (2017). The DigiSami project deals with the North Sami language (Davvisámegiela) which belongs to the Fenno-Ugric language family and is one of the nine Sami languages spoken in the northern part of Europe: Scandinavia, Finland and the Kola Peninsula in Russia (Seurujärvi-Kari et al. 1997). Figure 1 shows the Sami languages and their geographical distribution. There are about 40 000 speakers of the Sami languages, and of these about half speak North Sami, which has gained the status of a lingua franca. North Sami also enjoys official status in Norway and Finland, and …

Content truncated to 500 words