Short answers, pulled from the story.
In human-computer interaction, a modality is a single independent channel of input or output between a computer and a human. Channels can differ by sensory nature, such as visual versus auditory, or by processing type, such as text versus image.
A unimodal system has only one modality implemented, while a multimodal system has more than one. When multiple modalities are available for the same task, the system may be described as having overlapping or redundant modalities.
Vision can transmit information at 250 to 300 words per minute and audition at 150 to 160 words per minute. Tactition through a refreshable Braille display can average 125 words per minute, making vision and audition faster channels than touch.
The six types are equivalence, specialization, redundancy, complementarity, transfer, and concurrency. They describe different ways modalities combine or share information to convey it more effectively.
Complex human-to-computer modalities include computer vision, speech recognition, motion, and orientation. Speech recognition is widely used in virtual assistant applications, and motion and orientation sensors are common in smartphone mapping applications.
Vision and audition are most commonly used because they can transmit information at higher speeds than other modalities, at 250-300 and 150-160 words per minute respectively. Other senses such as taste, smell, and touch carry information more slowly and are rarely implemented.