A Longitudinal Bilingual Frisian-Dutch Radio Broadcast Database Designed for Code-Switching Research

A Longitudinal Bilingual Frisian-Dutch Radio Broadcast Database Designed for Code-Switching Research Emre Yılmaz1, Maaike Andringa2, Sigrid Kingma2, Jelske Dijkstra2, Frits van der Kuip2, Hans Van de Velde2,Frederik Kampstra3, Jouke Algra3, Henk van den Heuvel1 and David van Leeuwen1 1Centre for Language and Speech Technology (CLST), Radboud University, Nijmegen, Netherlands 2Fryske Akademy, Leeuwarden, Netherlands 3Omrop Fryslˆan, Leeuwarden, Netherlands {e.yilmaz, h.vandenheuvel, d.vanleeuwen}@let.ru.nl, {mandringa, skingma, jdijkstra, fvdkuip, hvandevelde}@fryske-akademy.nl, {frederik.kampstra, jouke.algra}@omropfryslan.nl Abstract We present a new speech database containing 18.5 hours of annotated radio broadcasts in the Frisian language. Frisian is mostly spoken in the province Fryslˆan and it is the second official language of the Netherlands. The recordings are collected from the archives of Omrop Fryslˆan, the regional public broadcaster of the province Fryslˆan. The database covers almost a 50-year time span. The native speakers of Frisian are mostly bilingual and often code-switch in daily conversations due to the extensive influence of the Dutch language. Considering the longitudinal and code-switching nature of the data, an appropriate annotation protocol has been designed and the data is manually annotated with the orthographic transcription, speaker identities, dialect information, code-switching details and background noise/music information. Keywords: Speech database, Frisian, code-switching, longitudinal data, radio broadcasts 1. Introduction Language contact has been extensively researched in the field of linguistics for more than 60 years (Weinreich, 1953; Auer, 1998; Muysken, 2000; Thomason, 2001). Contact induced language change shows up in the form of phono- logical, morphological, syntactic and lexical changes con- sequent to various linguistic phenomena such as word bor- rowing, interference and relexification. One prominent mechanism that is induced in the interacting languages is code-switching which is defined as the continuous alter- ation between two languages in a single conversation. Code-switching is highly noticeable in some minority lan- guages influenced by the majority language or majority languages that have been influenced by globally influen- tial languages such as English and French. Despite the well-established research line in linguistics, robustness of speech-to-text systems against code-switching and other kinds of language switches have recently received some in- terest (Yu et al., 2004; Lyu et al., 2006; Vu et al., 2012; Wu et al., 2015). Most of this work has been performed on Mandarin-English speech. In this work, we describe a novel speech database for the West Frisian language. West Frisian is one of the three Frisian languages (together with East and North Frisian spoken in Germany) and it has approximately half a mil- lion speakers mostly living in the province Fryslˆan located in the northwest of the Netherlands. The native speakers of West Frisian (Frisian henceforth) are mostly bilingual and often code-switch in daily conversations due to the exten- sive influence of the Dutch language (Popkema, 2013). The Frisian speech data has been collected in the scope of the FAME! (Frisian Audio Mining Enterprise) Project. This project aims to build a spoken document retrieval system for the disclosure of the archives of Omrop Fryslˆan (Frisian Broadcast) covering a large time span from 1950s to present and a wide variety of topics. …

Content truncated to 500 words