This study investigates contrastive hyperarticulation of voice onset time (VOT) in Japanese, focusing on voicing and length contrasts in spontaneous speech from the Corpus of Spontaneous Japanese (CSJ). Building on the assumption that lexical competition drives phonetically specific enhancement of phonemic contrasts (Aylett & Turk, 2004), we examine whether VOT —a key phonetic cue for voicing— is hyperarticulated when distinguishing minimal pairs. Target data (∼34 000 words) were extracted from 12 speech samples, balanced for speech style and produced by 11 speakers. Using phonetic and morphological annotations, we analyzed VOT duration normalized by speech rate, taking into account variables such as: presence/absence of minimal pair, place of articulation, position, word frequency, following vowel height and duration. We find that the presence of a voicing-based lexical competitor enhances VOT contrasts—shorter for voiced stops, longer for voiceless ones—while no such effect is observed for consonant length contrasts, where VOT is not a relevant cue. These findings support the view that language is shaped by communicative pressures: selectively enhance relevant phonetic cues in contexts of lexical ambiguity, thereby facilitating accurate message transmission. This cue-selective hyperarticulation reflects economy in speech production, where phonetic effort is deployed strategically to maximize communicative efficiency.
Sano et al. (Wed,) studied this question.