A Categorical Array is Crunch's term used to define a gathering of variables with the same categories. It produces a summary that looks like a grid. In the context of a survey, the respondent can select one option per sub-variable.
A typical Categorical Array is an agreement scale (eg: How much do you agree with each of the following statements? Strongly agree, Agree, Neither, Disagree, Strongly Disagree). But it could also be a numeric scale (eg: 0-10) represented categorically so that each of the values is a different category.
Below is an example where the categories are the frequency of newspaper readership. In the snippet below, we see 4 newspaper brands, so these would be stored in the data as 4 sub-variables that are bound together to make the categorical Array (each sub-variable has the same frequency categories in order from Almost always to Not Applicable.
In cross-tabs, researchers are used to seeing Column-%'s, but in a Categorical Array the default is to see Row-%'s, because you read left-to-right across each line (newspaper in the example above).
Note: these types of questions are different from the grid-style questions where you can select/check as many options as you like in a grid. In Crunch, that type of analysis is handled by Scorecards - where you line up different Multiple Response questions into a grid.