Working with Sources

Primary sources

Primary sources are documents produced at the time of the events by actors directly involved in the projects studied: architects, engineers, project owners, administrations, companies, users. These may include plans, but also letters, reports, technical surveys, construction site photographs, minutes, etc. They are mainly found in archives.
These sources offer direct access to the production logics of the built environment, but require a critical reading: a practice known as source criticism.

Secondary sources

Secondary sources are produced afterwards by researchers, historians, journalists, or other analysts. These include scholarly publications, articles, monographs, exhibition catalogues, or biographical narratives. They are primarily found in libraries or on digital platforms. They provide an interpretive reading, often valuable for framing a topic, but they must be confronted with primary sources to avoid uncritical repetition or simplification.

Scientific work involves articulating these two types of sources:

using primary sources as investigation materials, and secondary sources as framing or contextualization tools. It is in this tension that a grounded critical analysis is built.

Verifiable information, historically and materially situated

Verifiable information, historically and materially situated”,”desc1″:”The critical use of archival sources requires an awareness of the context of production, the producer’s intentions, and the relationships between documents that define any archival collection. The meaning of information is never intrinsic but is constructed through a situated reading process, taking into account materiality, provenance, initial function, and the document’s place in a network of historical, social, and institutional relationships.

Plausible discourse, generated on demand, unverifiable

Plausible discourse, generated on demand, unverifiable”,”desc2″:”AI does not seek information; it generates it. Language models (LLMs) such as ChatGPT do not directly access documentary sources, but generate responses based on probabilistic correlations between words: it is a statistical analysis of very large volumes of text. The “context” they evoke is a discursive simulation, built from recurring patterns, not the result of actual source research. The information provided is based on plausibility, not traceability: it is neither retrieved, nor situated, nor evaluated in its historical specificity.

For researchers, this implies methodological vigilance:

AI can serve as a linguistic tool or aid in formulation, but it in no way replaces the intellectual work of investigation, verification, and critical analysis of documented sources.

More information: Guidelines for the use of generative AI in research and education.

The objectivity of sources is often assumed. Yet every document is the product of a context and an intention. Sources do not simply reflect reality—they construct it according to specific logics: professional, political, aesthetic, editorial, or institutional:

• An architectural plan may be a working tool or a presentation document designed to convince a client.
• A text written by an administration or company may highlight political or economic interests.
• An architecture exhibition may construct a narrative that highlights some aspects of a project while obscuring others.
• A construction site photograph may be a simple stage record or an image intended to promote a work.
• An architect’s monograph may be a critical study or a self-promotion tool.
• An internal agency document has a different value than a published dossier.
• An old document may be reinterpreted differently depending on historical and cultural contexts.
• Some architects are overrepresented in architectural history due to editorial or academic choices.
• The historiography of modern architecture, for example, long privileged male figures and dominant movements, marginalizing other approaches.
• A plan does not show the entirety of a project: it privileges a normative and abstract point of view.
• An architectural photograph stages a building from an idealized angle, sometimes far from its real use.

All research must rely on a critical reading of sources. This means that documents (textual, graphic, audiovisual, digital…) should never be taken as neutral reflections of reality, but as situated data, linked to a context, an author, or a more or less defined intention anchored in its time of production.

Critiquing a source does not mean rejecting it: it means questioning its conditions of production, what it shows, what it omits, the filters or biases it may contain. It is essential to ask a few key questions:

• Who produced this document? (institution, author, commissioner)

• For what purpose? (communication, justification, documentation, internal use)

• When and in what context? (period, political, technical, or economic issues)

• What audience was it intended for?

• What does it leave in the dark? (absent actors, undocumented aspects, blind spots)

This critical stance allows tensions between sources to emerge and possibly identify contradictions or partial representations.

In the field of the built environment, where technical documents coexist with various narratives (from the architect, the engineer, public entities, clients, the press, etc.) or with regulatory frameworks, source criticism is essential to articulate the different dimensions of the project, which is at once material, political, symbolic, and social.

Adopting a critical stance ultimately means taking a position as a researcher, by assuming one’s methodological choices.

Proper citation of sources is a fundamental requirement of any scientific approach, not just an administrative formality. It not only acknowledges the work of institutions and individuals who produced or preserved the documents, but also ensures the traceability and verifiability of your work, giving it scientific weight. It shows that you rely on original, identifiable, and consultable documents for other researchers.

How to cite an archival source
An archival source is not a generic reference. It must be precisely identified by mentioning:

• the name of the institution (e.g.: Archives de la construction moderne, EPFL);
• the name of the fonds or collection (e.g.: Jean Tschumi Fonds);
• the complete reference code (e.g.: 0060.04.0160);
• the title or brief description of the document (e.g.: World Health Organization (WHO), Geneva: competition rendering [tracings]);
• the date (if known) and the author (architect, agency, etc.).

Example of citation:

• Archives de la construction moderne, EPFL. Jean Tschumi Fonds. 0060.04.0160 World Health Organization (WHO), Geneva: competition rendering [tracings], 1960. Jean Tschumi (architect).

And what about images?
When including an image in a scientific work (a reproduction of a plan, photograph, letter), it is not just a decorative illustration. An image is a document in its own right. It contributes to the argumentation and enables a critical reading of the project. As such, it must be captioned precisely, with the same information as for a written source: institution, fonds, reference, description, date, and author if available.

Note: mentioning the author of an image is not only a scientific requirement: it is also a legal obligation. Copyright law requires crediting all creations and, in case of publication or dissemination, requesting permission from the author or their rights holders.