Speech in 48 languages converges on intervals of about two seconds

The rhythm of conversation can seem to change sharply from one language to another. Syllables may be shorter or longer, and pauses occur in different places. Lara S. Burchardt, Ludger Paschen, and Susanne Fuchs asked whether that diversity remains when speech is viewed at a broader scale: the time from the onset of one vocal stretch to the onset of the next, including any pause between them.

The team analyzed more than 100,000 such intervals in recordings from DoReCo, a corpus of spontaneous, documented speech in many understudied languages. The sample covered 48 languages from every inhabited continent. The researchers calculated the duration and timing regularity of each sequence, then compared those measures with published data on fish, bird, and mammal vocalizations.

Although the distributions were not identical, typical durations in all 48 languages clustered around two seconds. The similarity appeared despite differences among language families and individual speakers. At the scale of whole stretches, the result indicates that languages vary less than they often do when rhythm is measured through shorter units such as syllables.

The cross-species comparison required normalizing duration because body size and other factors change the absolute timing of vocalizations. After that adjustment, human speech measures fell within the range reported for signals from the brown meagre, a fish, the skylark, and the Cape fur seal. The study does not claim that these animals organize messages like human language; it shows that the temporal distribution of their signals can occupy a similar range when analyzed with the same yardstick.

The authors propose two possible explanations for the roughly two-second pattern. Breathing limits how long a person can speak before interrupting the flow, while working memory may favor chunks that the brain can plan and track. Neither mechanism was directly manipulated in the study, so both remain hypotheses consistent with the data rather than demonstrated causes.

The convergence reframes the question of speech rhythm. Instead of looking only for a signature unique to each language, the work shows linguistic diversity operating within broader temporal constraints. Recordings from more languages and comparisons using equivalent protocols in other species could reveal whether the two-second scale mainly reflects breathing, cognitive processing, or a combination of both.