Working memory capacity has been suggested to depend on time-based duration of sounds over and above item-based number of chunks. Pronunciation speed and sound duration for numbers in Chinese (Cantonese) were shown to be faster than those in English. Digit spans in Chinese were also bigger than in English. There was a correlation between subjects' pronunciation speed for numbers and their digit span. Marginally significant correlations were also obtained between subjects' digit span and their mathematics grades. It is suggested that observed differences in digit span norms between language communities might be due to differences in pronunciation speed for numbers in the respective languages, and that the latter might also affect capacity for mental manipulation of numbers.