On the Limitations of Shape Grammars: Comments on Aaron Fleisher’s Article “Grammatical Architecture?”

Abstract

Shape grammars were introduced by Gips and Stiny in 1972. Since then, there have been many articles and books written by them and their associates. In 1992, Aaron Fleisher, a professor at the School of Planning, MIT, wrote a critique of their work in an article titled “Grammatical Architecture?” published in the journal Environment and Planning B. According to him, Gips, Stiny and later Mitchell, propose a hypothesis that states that shape grammars are presumed to represent knowledge of architectural form, that grammars are “formable,” and that there is a visual correspondence to verbal grammar. The strong version of “the hypothesis requires that an architectural form be equivalent to a grammar.” Fleisher considers these hypotheses unsustainable, and argues his case by analyzing the differences between language, and architecture, and by dealing with the concepts of lexicons, syntax and semantics. He concludes by stating that architectural design is negotiated in two modalities: the verbal and the visual, and that equivalences are not at issue; they do not exist. If there is such thing as a language for design, it would provide the means to maintain a discussion of the consequences in one mode, of the state and conditions of the other. Fleisher’s observations serve as the basis of this paper, a tribute to him, and also an opportunity to present an outline to an alternate approach or hypothesis to shape grammars, which is “nonlinguistic” but “generative,” in the sense that it uses production rules. A basic aspect of this hypothesis is that the only similarity between syntactic rules in language and some rules in architecture is that they are recursive.