The current study is dedicated to measuring vowel temporal acoustics (duration, durational difference, and durational ratio) in the medial position of mostly CVCVCV polysyllabic words in Arabic and Japanese, avoiding the asymmetries in vowel position, syllable structure, and coda consonant quantity (singleton versus geminate) observed in previous experiments. Twenty-nine (16 Arabic and 13 Japanese) participants were asked to use a carrier sentence to produce 60 polysyllabic (mainly CVCVCV) items that contrasted in vowel quantity (short versus long) and vowel quality (/a/, /i/, and /u/) at a normal speech rate. The results show that while short and long vowels are durationally distinct within a language, Japanese vowels are clearly longer than Arabic vowels, although the durational difference remains approximately the same between the two languages. The durational ratio of short-to-long vowel presents a new pattern that contrasts with that reported in earlier research. Specifically, Japanese long vowels in the medial position of polysyllabic words are twice as long as their short counterparts, while Arabic long vowels are more than twice as long. This shows that both vowel position and syllable structure must be considered when measuring vowel temporal acoustics or when structuring stimuli for perception experiments.

Previous research has utilized the duration ratio and occasionally the duration difference as single-value metrics to measure and compare the temporal acoustics of durationally contrastive vocalics (short vs. long vowels), which allow researchers to reduce two values (short and long) to one, but express a relationship instead of representing the vocalic duration values directly. The duration ratio may even be misleading when comparing two languages or dialects, as it is possible to exhibit a similar ratio but differ in durational acoustics, or vice versa. The current study proposes two alternative statistical metrics: a duration metric and a difference metric. The duration metric is an intermediate (mean-like) value between the duration of the short and long vocalics, and the difference metric is a ± value that can be added to or subtracted from the duration metric to obtain the duration of long or short vocalics. We conduct a production experiment on Arabic and Japanese vocalics and analyze the data using both traditional measures and the proposed metrics. The findings show that the proposed metrics better predict the language from which the vocalic duration values were obtained. Such results suggest that the proposed metrics are better candidates for measuring and comparing the temporal acoustics of vocalics.