For most people, interacting with a mobile device requires visual commitment to the input mechanism. As a consequence, there are many situations in our daily life when we have to refrain from using these devices, as our vision is already committed: for instance, while texting and walking in a crowded place, one should focus on the environment and not on the phone. A way to type fast, accurately, and with limited visual feedback is provided by chording keyboards. These devices generate a character by simultaneously pressing a combination of keys, and with five keys there are 31 combinations, enough for the letters of the English alphabet. If the keys are adequately placed, we can type with one hand and without looking at them. The main drawback of such keyboards is that the users should learn the correspondence between keys and characters before being able to type. In the first part of this thesis, we present a key-to-character mapping for five-keys chording keyboards and typing studies that estimate the achievable typing rates and error statistics. Besides this, we study the influence of having visual, audio, or no feedback at all on the typing process. The mapping, designed to minimize the learning phase, was completely learned after 45 minutes of training. Moreover, all participants said that this is a lot faster than they had initially expected. After approximately 350 minutes, the average entry speed is 20 words per minute with the maximum above 30 words per minute, regardless of the feedback type. Surprisingly at first, the character error rates are the lowest in the absence of feedback (2.32%) and the highest when one can see what has been typed (3.41%). We should also mention that the participants in the study were instructed not to correct eventual mistakes. Considering these results, the proposed text input method is a viable option for situations when vision is already committed to other tasks. In the second part, we focus on error correction. As we intend to use the keyboard without visual commitment, it is important to find a method to automatically correct typing mistakes. We present two such methods, specifically designed for a five-key chording keyboard. One central element of both methods are the probabilities that one character is typed for another. They are named confusion probabilities and were obtained experimentally. The first method is designed to correct single-word errors and is based on the maximum a posteriori probability rule. For the evaluation text, it reduces the error rate from 10.11% to 2.17%. As reference, MsWord and iSpell only reduce the error rate to 5.15% and 6.69%, respectively. The second method models the typing process as a hidden Markov process and considers the typing context, not only isolated words. In this way, the error rate is further reduced to 1.27%. This is more than three times lower than the two references. A very practical aspect of our work consisted in building several keyboard prototypes. For example, we designed, built, and tested a prototype for a bike. The five keys were placed under the natural position of the fingers on the handlebar. With the help of a wrapper application that captures the input text, we used the keyboard to control the operation of a smartphone while controlling the bike with both hands and staying focused on the road.