Eslo: From Transcription to Speakers' Personal Information Annotation

This paper presents the preliminary works to put online a French oral corpus and its transcription. This corpus is the Socio-Linguistic Survey in Orleans, realized in 1968. First, we numerized the corpus, then we handwritten transcribed it with the Transcriber software adding different tags about speakers, time, noise, etc. Each document (audio file and XML file of the transcription) was described by a set of metadata stored in an XML format to allow an easy consultation. Second, we added different levels of annotations, recognition of named entities and annotation of personal information about speakers. This two annotation tasks used the CasSys system of transducer cascades. We used and modified a first cascade to recognize named entities. Then we built a second cascade to annote the designating entities, i.e. information about the speaker. These second cascade parsed the named entity annotated corpus. The objective is to locate information about the speaker and, also, what kind of information can designate him/her. These two cascades was evaluated with precision and recall measures. 1. Corpus presentation The most extensive examination of spoken French before 1980 is the Socio-Linguistic Survey in Orleans (Enquête Socio-Linguistique à Orléans, ESLO 1). This investigation was carried out towards the end of the Sixties by British academics with a didactic aim and represents a collection of 200 interviews with several references (sociological characterization of the interviewed persons, identification of the interviewer, date and place of the interview). The interview was recorded in a professional or private context, in a total of 300 hours of speech and the corpus contains approximately 4,500,000 words. Note that a new survey, ESLO 2, has been undertaken by the LLL-Orléans in order to constitute a corpus comparable in terms of data gathering and archiving. ESLO 1 and ESLO 2 will form a collection of 700 hours of recording, that is more than 10 000 000 of words. In this article we want to show the different levels of corpus annotations. Starting from the transcription which, we believe, is the first level of annotation, we will then describe how to increase the transcription with some semantic annotations in XML format. 2. Preliminary work on ESLO The ESLO project contains many steps: numerization, coding of metadata, transcription synchronic with the sound, annotation, anonymization, tools of request and diffusion. The main theoretical and technical choices operated during the scientific exploitation of the corpus ESLO1 answer to a precise objective: to participate in the reflection on the evolution of models and methods of constitution and exploitation of the oral corpus for linguistic purposes [Abouda, Baude, 2007]. 1 The part of ESLO1 was put online (http://bach.arts.kuleuven.be/elicop/) within the Elicop project [Mertens, 2002]. ESLO1 was consisted of the wave band of recordings, index card of sociological characteristics, descriptions of situation (date, place, remarks on the acoustic, etc.). Audio files were digitized; an indexation and a first cataloguing were accomplished. The objective is to make available all data to the scientific community in a format which allows optimal and intensive exploitation [Baude, 2006].