Falling silent, lost for words ... Tracing personal involvement in interviews with Dutch war veterans

Falling silent, lost for words ... Tracing personal involvement in interviews with Dutch war veterans Henk van den Heuvel, Nelleke Oostdijk CLS/CLST, Radboud University Erasmusplein 1 Nijmegen, the Netherlands E-mail: {H.vandenHeuvel|N.Oostdijk}@let.ru.nl Abstract In sources used in oral history research (such as interviews with eye witnesses), passages where the degree of personal emotional involvement is found to be high can be of particular interest, as these may give insight into how historical events were experienced, and what moral dilemmas and psychological or religious struggles were encountered. In a pilot study involving a large corpus of interview recordings with Dutch war veterans, we have investigated if it is possible to develop a method for automatically identifying those passages where the degree of personal emotional involvement is high. The method is based on the automatic detection of exceptionally large silences and filled pause segments (using Automatic Speech Recognition), and cues taken from specific n-grams. The first results appear to be encouraging enough for further elaboration of the method. Keywords: discourse annotation, emotion recognition, speech databases, automatic speech recognition, oral history 1. Introduction Eye witness reports that relate first hand observations and personal experiences of events happening do not only include factual statements about ‘who did what to whom’ or ‘what happened when’, but also the witness’ opinion or attitude towards what he or she has witnessed or experienced. Witnesses in their accounts of one and the same event will differ in a variety ways, for example in their choice of words or the attention they give to certain details by elaborating upon them. Combining what information is contained in multiple reports, can help the reconstruction of what (may have) happened and how witnesses say what they think of it. Where a witness report has been captured and kept as an audio recording, there is important additional information that can be learned from the way that things were said, that are not conveyed through the wording, but by making use of various paralinguistic means (prosody, pitch, volume, intonation). Thus, in spontaneous speech, silences and filled pauses (like 'eh') are important cues to a speaker's emotional state and involvement (Tisljar-Szabó & Pléh, 2014). In sources used in oral history research (such as interviews with eye witnesses), passages where apparently speakers personally are very much emotionally involved are of particular interest, as these may give insight into how historical events were experienced, and what moral dilemmas and psychological or religious struggles were involved (Van den Berg et al., 2010). So far, however, studies in oral history have mostly been based on written versions (verbatim transcriptions), rather than on the primary audio data, as exploration of the audio is not trivial. In the present paper we report on a pilot study we conducted. Using a large corpus of interview recordings with Dutch veterans (IPNV; Van den Heuvel et al., 2012) we investigated if it is possible to develop a method for automatically identifying passages where the degree of personal emotional involvement is high. 1 The data originally …

Content truncated to 500 words