Another type of integration task asks for similarities and differences within a list or between lists. Note that here we are talking about the basic ‘list structures’ found in documents. Graphs, for example, represent lists of values. The ‘graphic’ representation of these lists make integration tasks easier. In the ‘required hours’ table shown in the example we could ask about the maximum and minimum number of hours required to qualify for EI, an integration task that requires comparison of all the amounts in the right hand column. Of the same table we could ask, “What is the relationship between regional rate of unemployment and the hours that must be worked to qualify for EI?” Answering this question would require a comparison of the ordering of the lists in the two columns, a fairly high-level integration task. (Note that the ‘type of information,’ a pattern or relationship, is also quite abstract.)

Generating requires readers to use expert background knowledge to relate information in the question or directive to the text, or to select one plausible answer over another.

When searching for information, we are often confronted with situations where the term we are using as given information fails to match any of the available categories or labels in the document we are using. Kirsch and Mosenthal call this situation ‘coping without cues or clues.’ For example, imagine you were looking for the phone number of a business that sold ‘eyeglass frames.’ A search of the yellow pages reveals that there is no listing under ‘eyeglass’ or ‘eyeglass frames.’ It is evident that if the search is to continue, a more productive search term will have to be ‘generated.’ Most people know that eyeglass frames are sold by ‘dispensing opticians’ will regard this task as fairly trivial; for other Canadians, this is a piece of ‘expert knowledge’ they do not have.

We have only two strategies for ‘generating’ search terms. We can ask someone who knows or we can ‘brainstorm’ some possibilities based on linguistic, semantic, and logical cues. The former is apparently what male drivers should do more often, the latter is a high-level cognitive activity of which a substantial number of people are incapable.

A lack of congruence between how information is framed by the ‘information designer’ and how it is perceived by the ‘information consumer’ results in poor communication. Examples of this lack of congruence abound in all sorts of public documents. In Calgary, the ‘Animal Control Bylaw’ covers ‘horses, dogs and fowl.’ If you are bothered by cats, you will need to refer to the ‘Cats Running at Large Bylaw.’ If you are looking for an industrial backhoe, the Yellow Pages for one city categorizes it as ‘equipment,’ another as ‘machinery.’

Some category generation is also required when requests for information fail to specify the kind of information required, or the basis upon which distinctions need to be made. When categories are not provided, the respondent has to ‘generate’ relevant categories for the information he or she has. For example, the description of the person who robbed the store will be more detailed if the witness does not have to ‘generate’ the categories which are used in the description. An outline of categories such as ‘height,’ ‘build,’ distinguishing characteristics,’ etc. is generally provided to make this task easier.