Many textbooks say that strong sounds in English come at equal times, like a clock. But researchers measured real speech, and the times were not equal.

Even so, English and Japanese sound different. In Japanese, most beats have a vowel and are about the same length. English has groups of consonants, like in "strengths," and many weak, short vowels. So English sounds bumpy, and Japanese sounds smooth. To sound more natural in English, make the weak parts short.

English is often described as "stress-timed," meaning that stressed syllables come at regular intervals, while Japanese is called "mora-timed," with beats of equal length. However, when the phonetician Dauer analyzed real speech in 1983, the time between stresses in English became longer when there were more syllables, so the intervals were not even.

Even so, English and Japanese clearly sound different. In 1999, Ramus and his colleagues measured eight languages and found that Japanese had the highest share of vowels, while English had less vowel time and more varied groups of consonants. English also has many reduced vowels, which are weak and short. This explains why English sounds uneven to Japanese ears, and why learners should keep weak syllables short.

Students are commonly taught that English is stress-timed, with stressed syllables recurring at roughly equal intervals, whereas Japanese is mora-timed. Acoustic measurements, however, have not supported the strong version of this claim. Analyzing English alongside languages such as Spanish and Italian, Dauer found in 1983 that the interval between stresses grew with the number of syllables and that its distribution in English was not very different from that in so-called syllable-timed languages. She proposed that perceived rhythm arises instead from syllable structure and vowel reduction.

Ramus and colleagues tested this idea in 1999 by dividing recordings from eight languages into vocalic and consonantal intervals. Two measures, the proportion of vowel time and the variability of consonant intervals, separated English and Dutch from Japanese, which showed the highest vowel proportion. Perception studies by Otake, Cutler and others also suggest that Japanese listeners segment speech into morae. That said, these measures are sensitive to speaking rate and were based on only a few speakers per language, and many researchers now see rhythm classes as a continuum rather than strict categories.