Building CLIA for Resource-Scarce African Languages: A Case Study on Oromo—English CLIR

Introduction

As we move towards an increasingly globalized and knowledge-based economy, the ability to instantly access and share relevant information (Baeza-Yates & Ribeiro-Neto, 1999; Gey, Kando, & Peters, 2005; Nie, 2010) beyond language and cultural boundaries has become more and more crucial. The World Wide Web (WWW) contains massive volumes of multilingual and multimedia information resources that can be explored and exploited to address critical social and economic problems. Unfortunately, in developing and culturally diverse regions like Africa and Asia, the accessibility and usability of online resources are severely constrained by formidable obstacles and challenges such as language barriers, linguistic digital divide and lack of robust CLIA systems (Adegbola, 2009; Gasser, 2006; Varma, Tune, & Pingali, 2007). As pointed out by (Georg & Hans, ‎2013; Oard & Diekema, 1998; Peters, Braschler, & Clough, 2012), language barriers and linguistic digital divide have continued to threaten and undermine the potential of the Internet to deliver universal and equitable access to online information resources and services. This is especially true in highly multicultural developing nations like Ethiopia and India.

Broadly speaking, language barriers can be defined as linguistic and cultural factors that impede the free flow of information across language boundaries. In this article, the term language barriers is more specifically used to describe linguistic and cultural obstacles that discourage or prevent users from seeking and sharing important information across different languages and cultures. Even though the term linguistic digital divide is closely associated with language barriers, it is often used to describe the disparity in technological development between different languages (Gasser, 2006; Scannell, 2007). While the term digital divide is generally used to describe the gap in accessing and using computing devices among various social groups, the term linguistic digital divide is more specifically used to describe the relative advantages of certain languages (or language communities) over the others with respect to modern language resources and information access technologies.

Since most of the existing commercial search engines and Information Retrieval (IR) systems have primarily focused on well-resourced European and Asian languages, they have not paid adequate attention to supporting under-resourced African languages (Adegbola, 2009; Gey, Kando, & Peters, 2005; Osborn, 2010; Pingali, Tune, & Varma, 2008). The need for exploring and developing multilingual information access technologies that permit African communities to search and discover information beyond linguistic and cultural barriers has, therefore, become more urgent today than ever before. In this regard, much attention has been paid to the development of Cross-Language Information Retrieval (CLIR), which is mainly concerned with searching and discovering information beyond language and cultural boundaries (Hedlund, et al., 2004; Nie, 2010). The main purpose of CLIR is to identify documents written in one or more language(s) in response to a query expressed in a different language (Nie, 2010; Peters, Braschler, & Clough, 2012). On the other hand, CLIA deals with much more general and broader issues. CLIA encompasses not only the academic domain of cross-language search or CLIR, but also many aspects of natural language processing and understanding, including text encoding, digitization, content analysis and visualization (Peters, Braschler, & Clough, 2012). In this paper, we use the term CLIA in its narrower sense to refer to the processes of querying, accessing and retrieving information across different languages.