We know that "people often pronounce a word differently when asked to read written material aloud than when speaking spontaneously." We also know that computer developers usually teach computers that recognize speech "by using samples of written material read by the people who will be using the computers."
As the computers are trained using written material read aloud and people usually read words differently when reading written material aloud as compared to when speaking spontaneously, it is likely that computers will have a harder time handling spontaneous speech than it will written material that is read aloud. This is because they are already familiar with the pronunciation of words read from written material.
This is directly restated by answer choice C, making it the correct answer.
Hope this helps! Please let us know if you have any further questions.