The phonetic reduction of nasals and voiced stops in Japanese across speech styles

In daily conversations, speakers often produce speech in a casual manner. Casual speech, also referred to as spontaneous or conversational speech, contains a high degree of variation as compared to more careful speech styles (Ernestus and Warner, 2011). One important aspect of casual speech leading to this high variability is phonetic reduction, resulting in words being pronounced with fewer segments, shorter durations, and assimilation. For example, yesterday pronounced carefully could be something like /jEst@~deI/ but in casual speech it could be pronounced [jESeI] (Tucker, 2007). Reduced pronunciation variants has been studied cross-linguistically with evidence being reported in American English (e.g., Johnson, 2004; Warner and Tucker, 2011), Dutch (e.g., Ernestus et al., 2002), French (e.g., Brand and Ernestus, 2015), Finnish (e.g., Lennes et al., 2001), German (e.g., Kohler, 1990), and Japanese (e.g., Arai et al., 2007; Maekawa, 2005). In the present study, we use a large-scale speech corpus, the Corpus of Spontaneous Japanese (Maekawa, 2003), to examine the phonetic variability found in nasals and voiced stops and to describe how that variation and reduction occurs across speech styles in Japanese. Using the Corpus of Spontaneous Japanese, we analyzed the duration and intensity difference of target segments across four styles of speech: academic presentations, simulated public speech, dialogues, and read speech. The intensity difference was defined as the difference between the minimum intensity of the target segment to the averaged maximum intensity of surrounding segments (Tucker, 2011; Warner and Tucker, 2011). We hypothesized that we would observe stronger reduction (more approximant-like productions), as indicated by shorter duration and smaller intensity difference, as speech style becomes more casual. In other words, the shortest duration and the smallest intensity difference would be found for nasals and voiced stops in dialogues (most casual) and the longest duration and the largest intensity difference in read speech (least casual).