In simultaneous interpreting, the gap between what the interpreter hears and says is called the ear-voice span: usually two to three seconds, sometimes more. Too short and you risk mistranslating; too long and you forget. A classic headache is German, where the verb often comes at the end of the sentence, forcing interpreters to anticipate or wait.
In Kuşköy, a mountain village in northern Turkey, people talk across valleys by whistling Turkish. A 2015 study showed that whistled Turkish draws on both hemispheres fairly evenly, whereas spoken Turkish relies mostly on the left. The physical form of the signal, it turns out, matters to the brain.
Speech has no gaps between words. In 1996, researchers played 8-month-old babies a continuous stream of made-up syllables for just two minutes. The infants picked out the 'words' by tracking which syllables tend to follow one another. That's probability calculation, done without a calculator or any visible effort.