We will be publishing Nikos Salingaros’ book, Unified Architectural Theory, in a series of installments, making it digitally, freely available for students and architects around the world. The following chapter discusses the complexity of form languages and describes how to use the form language checklist to measure these complexities. If you missed them, make sure to read the introduction, Chapter One, Chapters 2A and Chapter 2B, and Chapter 3 first.
There exists a volume of writings by architects in the early 20th century, and we can look through them for the form languages of Modernism. Unfortunately, the useful material turns out to be very little, most of it describing not a form language but rather marketing and declarations of a political nature. Moreover, those pieces of very personal form languages are presented as normative theories: a prescription of what to do and what not to do, with the weight of universal ethics, even though they are based solely on opinion, not empirical observations or systematic study.
Here are some practical lists of rules I have found by Naum Gabo and Antoine Pevsner, Ludwig Mies van der Rohe, and Le Corbusier.
By the brothers Naum Gabo and Antoine Pevsner, 1920: “Reject closed mass and volume, and model space from within outwards. Reject color, and use only the natural color of the building materials. Reject all ornament.”
Ludwig Mies van der Rohe, 1923: “Open plan for interiors. Materials are limited to concrete, iron, and glass. Use only curtain walls and reinforced concrete — no load-bearing construction.”
Le Corbusier, 1927: “Lift the building from sitting with its basement in the earth, to being suspended on posts (pilotis). Only curtain-wall construction is allowed. Roofs have to be flat. Windows can only be horizontal and will extend from one load-bearing pillar to another, which makes them very wide (narrow and long).”
These three sets of rules for a modernist form language do contrast with traditional form languages, so that of course the product looks markedly different from a traditional building erected prior to the 20th Century. This “new look” was part of the modernist form language’s appeal when it was first introduced.
Without discussing the merits of the modernist form language and its variants here, there is an appeal to universality, and thus a rejection of regional adaptation. There is also a strong motivation to reject elements simply because they belong to traditional form languages: turning against one’s cultural tradition for the sake of innovation.
Our society adopted the modernist form language for an enormous number of buildings, and so we have forgotten the store of older and traditional form languages. This represents a great loss for the knowledge base of the architectural profession. No rational society should throw out practical information, unless that body of knowledge has been proved wrong, or is no longer useful.
Nothing wrong was ever discovered in older form languages: indeed, they have a great number of adaptive qualities that create pleasant, functional, and comfortable living and working environments. We suggest that an architect can learn from all form languages. Some languages are going to be more relevant to the location than others — a welcome return to valuing regionalism because this leads to sustainability.
When deciding to employ an older form language to design a building today, the architect has an option. He or she may use the form language in its original form. Otherwise, the architect may choose to upgrade it by introducing improvements or savings through more contemporary materials. Architects also have the option to add individual innovative elements of their own, unless commissioned to design in a particular form language.
A form language evolves in time, just as a written and spoken language does, so change is natural. What is not natural is drastic reversal in a form language. The crucial concern here is to modify an architectural form language so that it does not lose its adaptive and expressive power. In order to achieve this, the architect must begin from a deep respect of what evolved traditional form languages represent.
An in-depth study and analysis of a particular form language prepares you to use it as a design tool by understanding how a design arises from the combinatoric “linguistic” structure of forms. If the student properly and accurately documents a form language, then it could be used to design an entirely new building. The measure of success is if an observer thinks that the new design resembles an original building enough to be considered as arising from the same language. I want to end the common practice of students copying buildings as images directly, which is both unintelligent and uncreative. The proper way to design in a particular language of choice is first to extract and document that form language from one or more examples, then use the form language to design a new building.
Documenting a Form Language
Going through the process of documenting a form language is an educational experience. First, it reveals the complexity of the language: how many words (and diagrams) are required to describe it so it can be applied to design something. There exists a very simple measure of complexity that we can employ here. The Kolmogorov-Chaitin complexity measure is the minimum length of a system’s descriptor. It’s the “length of code” without redundancies. For a form language, it would be the word count of your completed “form language checklist” [template provided later in this book].
This first measure of the complexity of a form language opens up new dimensions for understanding architecture. Satisfying user needs, adaptations to climate, region, and materials should make a form language more complex — with a longer word count. Actually, both highly-ordered and random systems are complex, but in a different way. We will study that distinction later on. Now we note that complexity of form language does not necessarily imply adaptation, and will look for a correlation between Kolmogorov-Chaitin complexity and regional adaptation.
The model also allows us to compare very different form languages in terms of their complexity. Distinct form languages cannot be compared visually, because of the very different images they present, but rather in terms of each language’s overall complexity.
Traditional regionalism involves adaptation to local materials, climate, culture, and societal practices. (We will discuss later the possibilities of combining regionalism with 20th-century modernism.) Using the model of measuring the complexity of a form language through a word count of its verbal description, we can investigate how adaptation to local practices and building culture requires a longer or shorter description of the design process. Our intuitive experience would lead us to say that better adaptation to local requirements requires a longer description.
A form language is one prescription for creating structural order, and its products will have their own characteristic appearance. At the heart of every spoken or written language is a set of rules common to all languages. We can look for these general rules in other sciences to understand the commonality among visually distinct architectural styles.
I introduced some rules for structural order to help explain Alexander’s theory of adaptive design, which we study later. These rules are taken from physics, not architecture, and establish a helpful rubric for analyzing form languages. They represent the means of achieving coherence of forms.
Three laws for architecture are proposed: (1) Smallest-scale order consists of paired contrasting elements. (2) Large-scale order occurs when every element collaborates to reduce randomness. (3) Small is connected to large through a hierarchy of intermediate scales, using a scaling ratio around e ≈ 2.7.
Let’s see the consequences of these rules for creating coherent order. The smallest scale will need to have well-defined components so that they can indeed couple. It cannot be empty altogether. Coupling is achieved through geometric interlock and contrast. It then follows that whenever there is repetition, it is a coupled pair that repeats.
Randomness is reduced by using symmetries of all types: repetition, alignment or translational symmetry, reflectional, rotational, and glide symmetries (which are a translation plus a reflection). The intention is to be able to experience the structure as a whole, rather than to have to account for each individual component separately. Components on the same scale are related using common symmetries, whereas those on different scales are related through scaling symmetries.
These rules, originally given in the context of a specific theory of design, prove useful in writing down form languages. For example, look for repeating and contrasting (paired) components on the small scales. Pay attention to what is happening at many distinct scales. Look for symmetries, or their absence where they would ordinarily be expected.
We are also sensitized by this framework to recognize frames and boundaries in form languages. In traditional building and design, two structural elements rarely come together without some form of trim, intermediate region, or border. This was eliminated in the minimalist form language we are most familiar with, so we should not overlook the boundaries — which appear in all traditional form languages — now.
Non-adaptive form languages could in fact be generated by rather simple rules. For example, “generate a sculptural form with a computer program then build it as a building”, or “crumple a piece of paper then build it as a building”, or “draw a doodle on a piece of paper then build it as a building”. These are descriptions of only a few words. Yet they rely upon some industrial form language, say, for warehouses or aircraft hangars, to get the job done — the sculptural model is not enough to make working drawings for the contractor. Brief rules like these work together with a developed form language. The result is a building having a very complex description.
Another brief rule that can generate a complex form language works by reversing or negating an existing form language. Again, this requires a developed language to act upon. We can imagine prescriptions such as: “reverse the scaling hierarchies”, or “eliminate straight lines”, or “smash forms to the point just before they become uninhabitable”, or “slash walls to cut diagonal strip windows”. These simple rules change an existing form language radically, and create complex buildings with novelty appeal. Otherwise, an architect can drastically simplify an existing form language with the rule “strip everything off except supporting structure”, which reduces its complexity.
Architects are not in the habit of writing down their form languages. Either they feel they own a design secret that they don’t wish to see copied by others, or they are simply not used to documenting design in this manner. It could also be the case that they are generating their form language by a “shortcut”, such as one of the above. Other architects and researchers usually study buildings afterwards, and these are the people who delve into a form language. But even those don’t usually document the form language.
Measuring complexity using the form language checklist
Please just fill out the details for your chosen form language as succinctly as possible in the checklist below. The entries include lists of materials, forms, sizes, etc. For some entries, a simple yes or no indicates whether something (a structural element, or property of that element) is present or not. It will be necessary to estimate both actual sizes of components, and relative ratios of sizes among different components. In listing the connections, an unusual element of interest (at least for today’s design thinking) is to look for and document an intermediate piece that connects two other components. In many contemporary buildings, this intermediate connection is missing for stylistic reasons, so to perform this exercise you will have to change the way you look at structures.
Then, please compute the Kolmogorov-Chaitin complexity of your form language by using the word count of your completed checklist. The more “wordy” your checklist, the more complex is the form language. Also roughly estimate the regional adaptation of your form language on a scale of 0 to 10 (higher for better adaptation). This is the simplest possible estimate of regional adaptivity of your building, which represents the opposite of any abstract, formal, or “universal” design method. An “International Style” building will necessarily rank very low. Finally, look for any correlation.
What is remarkable is that we are able to measure the complexity of a form language at all, and by a simple means such as the word count by a word processor. We now have the beginnings of a new investigation of complexity in architecture. We will go further and correlate this complexity with adaptability and regionalism much better, in our second model presented later.
Form Language Checklist
NAME OF FORM LANGUAGE: location, era, name of architect, particular building?
DOCUMENTATION: is there a written description or set of working rules for this form language? (Instructions, not a philosophical or ideological justification).
MATERIALS: titanium, steel, glass, brick, concrete, wood, stone, adobe, thatch, etc.
COMPONENTS: walls, floors, roofs, beams, windows, doors, and their dimensions.
CONNECTIONS: cornices, joins, moldings, meeting points of wall+wall, wall+floor, wall+window, door+wall, wall+ceiling, façade+roof, size of connection compared to what it joins.
OVERHANGS AND CANTILEVERS: type of supports, placed on top or bottom?
ARCHES: yes/no, type, spacing, height, dimensions.
COLUMNS: yes/no, type, size, width, alignment, inter-columnar spacing, fluting?
COLUMN CONNECTIONS: column+floor = base, column+top = capital, relative size.
RECTANGULAR OR OTHER GEOMETRY: rectangular, diagonal, or curved.
CHARACTERISTIC FORMAL SHAPES: overall geometry of components, their relative alignment, and their variety.
SUBDIVISIONS OF FORMS: yes/no, for walls, for windows, their relative dimensions.
GRAMMAR AND SYNTAX: what components relate to each other (symmetry), or should not relate to each other (asymmetry). Any hidden rules?
ENTRANCE: relative size to other components, method of definition, change of level?
PORCHES AND BALCONIES: yes/no, depth, roof connections, front grill or solid?
FLOOR PLAN: subdivision of space, order and hierarchy of rooms, circulation.
EXISTENCE OF SCALES: well-defined and usually repeating structure on 1mm, 3mm, 1cm, 3cm or 1in, 10cm, 1m or 1yard, 3m, 10m, and other scales.
COLOR: yes/no, which ones? Intensity? Do different colors harmonize?
LARGER SYMMETRIES: formal symmetries on scale of 10m down to about 1m.
SMALLER SYMMETRIES: sub-symmetries from 1m down to fine detail.
DECORATIVE ELEMENTS: non-functional large elements used only for style.
ORNAMENT: yes/no, type and design, scales on which it appears, extent.
SURFACES: materials and textures presented to user, “friendly” or not?
Architectural regionalism correlates with design complexity
After documenting a particular form language, we measure its “complexity” by using the word count of the description according to the completed “Form Language Checklist”, above.
At the same time, we estimate the regional adaptation on a scale from 0 to 10, with 0 being the least adaptive to locality, building culture, and specific user needs tied to local culture. We are guided in this estimation by what regionalism means in the context of using local materials, employing traditional typologies, low-cost methods of energy use and optimization, historical continuity of design typologies and the use of traditional ornamentation, etc.
Each form language presents an ordered pair of numbers (word count, regional adaptation), and these can all be compared. It is most instructive to plot Regional Adaptation (on a vertical axis) versus Complexity of a Form Language as measured by the word count of its verbal description (on the horizontal axis).
Such a plot does indicate that the regional adaptation correlates with the complexity of a form language. This result is all the more striking because a measure of form language complexity obviously depends on the word count, and is also dependent upon the verbosity of the person describing it! Despite the evident inaccuracies of the method, these results open a very promising topic for more detailed investigation.
During class discussion evaluating each building in the context of all the results, several students said that their choice of building “turned out not to be a good example after all”. Pressed to explain this, they stated that they were originally attracted to their building because of the usual architectural design criteria, but our analysis showed them that more important qualities facilitating human use and simple economics were lacking. As a result, they understood how to judge whether a form language was adaptive or not.
Further Reading:
Nikos Salingaros, A Theory of Architecture, Chapter 1, “The Laws of Architecture From a Physicist’s Perspective” (Umbau-Verlag, Solingen, 2006).
Nikos Salingaros, “Kolmogorov-Chaitin Complexity”, Meandering Through Mathematics, 23 September 2012. Chapter 9 of Unified Architectural Theory.
Nikos Salingaros & Kenneth Masden, “Against Ecophobia”, Philadelphia Society, 8 October 2011. Also available in GREEK and ITALIAN. Chapter 10 of Unified Architectural Theory, and we will be publishing this essay online.
Henry Glassiem “Folk Housing in Middle Virginia,” 1975.
Order the International edition of Unified Architectural Theory here, and the US edition here.