How engineering designers talk

In a recent study, my colleague Dr. Lucian Ungureanu analyzed the details of how engineering designers communicate with each other. Lucian closely looked at the language designers used in two recorded design meetings that were concerned with the design of a crematorium . To this end, he applied a text mining technique called n-grams, that identify sequences of words (with length n) that are frequently used within a specific communication event. For example, during the two design meetings of the crematorium, participants used the 4-gram ‘it would be nice’ 17 times.

Looking at the n-grams Lucian found that many of the most frequently used ones, are used by designers to express ambiguity and vagueness. For example, ambiguous and vague statements. For example, the 6-gram ‘it would be nice to have’ points towards a vague statement as it also opens possibilities to alternative choices. Instances of vague 2-grams would be, for example, ‘bit bigger’ or ‘additional space. Both of these 2-grams do only point towards an extension, but do not point to the precise magnitude of the extension.

One conclusion that can be drawn from the relative frequency of ambiguous and vague n-grams is that designers use such ambiguous and vague phrases to indicate important ‘design acts’ – statements and discussions within meetings that lead to crucial design changes or new innovative design ideas. The findings also point to the overall importance of ambiguous and vague statements during the evolution of design ideas and changes. They allow designers to suggest new avenues and angles without necessary specifying specific solutions already and invite others into the discussion.

Thinking about computational design tools, this research can provide many insights. Most importantly, the results point towards the need of computational design support tools that allow designers to include ambiguous and vague ideas – something that is, by large, not possible with existing tools yet. At the same time, it is probably possible to develop intelligent support agents that can detect ambiguous and vague statements and then provide dedicated design support during meetings, such as triggering specific visualizations or additional supporting information and data, on the fly, within meetings.

If you are interested in this research you can find the full paper here.