Context:

This installation, an iteration of a wider collaboration between Natalia Janula and Konstantinos Damianakis, emerged through a site-specific exploration of River Thames during the interdisciplinary CoLab festival in Trinity Laban 2026. The two artists lead a week-long workshop for a small group of music and dance students of the Conservatoire, exploring confluenceas a material metaphor for collaboration. Confluence, inspired largely by hydrofeminist theory, as articulated by Astreida Neimanis – a kind of “collective worlding” between differing “bodies of water”, including “with other bodies and times” (29, 4) – allowed for the immanent emergence of the installation, by amplifying the common flows between different members of the group, spaces, material, and stories involved. These materialised in an immersive interactive installation in Loggia and Philip Carne Room, within Kings Charles Court in Greenwich, on the south bank of the Thames.

Text:

Text was diffused throughout the process, approached through the notion of ‘wavewriting’, developed by Andreas Philippopoulos-Mihalopoulos – another collaborator of the wider project – in his recent book, Hydrojustice (2025): “writing the laps and ebbs without synthesis, letting the meaning be adumbrated rather than declared, the nuances to flood the ideas, the new emergences to be absorbed by the horizontality of the surface and the circularity of the return” (p. 27). It was neither used as one-way instructions or a theory to practice, nor as a way of simply describing or reflecting on what happened during the practice. The agency of the text rather flowed along with the rest of the material, circling between the role of a text score, the puddling of embodied experience, a kind of Neimanian ‘proxy story’ amplifying and sensitising the sensations encountered, as poetic offerings for the birthing of the world we were making.

Each day begun by introducing a textual fragment to the group: an excerpt by the ‘Listening’ chapter of Alexis Pauline Gumbs’ Undrowned: Black feminist lessons from marine mammals (2020), the section titled ‘Reading a Wave’ from Italo Calvino’s Mr. Palomar, and a part of the afterword to Rivers Solomon’s The Deep by clipping. (2019), covering the three practical dimensions of ‘listening’, ‘scanning’, ‘interacting’. Each text was followed by a day-long embodied engagement with the themes of each text, as described later. Some of the activities also included moments of shared writing, almost like transcribing the various encounters, the feelings that emerged, or various imaginings.

The emergent linguistic soup (a term misappropriated by Manuel DeLanda, but fitting for the way it portraits language as the shared confluence of sounds and symbols in a collective assemblage), flowed visually through the installation. By accumulating and analysing the various excerpts, poems, messages, and stories, we created a latent space of letter-to-letter-transition probabilities, within a Markov-chain-based (wave)writing machine.

The choice of Markov chains is informed by Konstantinos’ PhD methodology, and the way he works with the notion of the machine in the work of Deleuze and Guattari. The Deleuzo-Guattarian machine, differs wildly from other common versions of the concept. It neither refers pragmatically to a device, or a technical object, nor to the progressive capture of previous forms of labour, as in the works of Babbage and Marx. In short, expanding Ada Lovelace’s extension of machinic realm from the numerical to the symbolic (Pasquinelli 2023: 71), the machine of Deleuze and Guattari extends the symbolic towards the real, it concerns the radical openness and interconnectedness of heterogeneities entities, both actual and virtual (Savaugnargues 2016: 185-218; Raunig 2010: 26-34). Markov chains, according to Anne Sauvagnargues, is part of the scientific ground for the philosophers’ interest in a-/pre-signification, since they embody the possibility of the production of language and meaning ‘by taking into account only real occurrences and not their significations’ (207). Markov chains are opposed to a ‘thematic development’ as their modus operandi is an ‘open formation’, accessing ‘the mode by which forms communicate’ (Sauvagnargues, 207), resonating with what Andreas Philippopoulos-Mihalopoulos describes as the aqueous ‘depth of the surface’ (p. 23), a textual surface, in our case, as an open formation ‘where the depths rise and mingle’ (22). In Stay Hydrated, this textual surface keeps flowing, leaking out new symbols every time someone takes a step within the room (with the use of a geophone and a specifically tuned trigger detection algorithm), and mingles with other surfaces: the glasses in the aquarium, the water surface which turbulently reflects it on the walls of the rooms and the glass doors that in turn reflect in on other surfaces here and there, and so on.

Sound:

A concept that guided our approach to sound, was listening confluently, as it previously emerged in the research of Konstantinos, particularly in a sociology project on the place-making around the rivers of Lewisham (London), which culminated in two podcasts and an audio essay titled Listening for Mermaids: The Otherworldly Life of Lewisham’s Rivers(Copenhagen: Seismograf, 2025). In short, this concept plays with the figuration of a river confluence to understand ‘listening’ as the point of convergence of different ‘sound streams’, and vice-versa, with an emphasis on the posthuman vectors that appear in the corresponding imaginaries and practices.

During the workshops, we embarked on soundwalks along the banks of River Thames during the low tide. With us, we had a recorded, two hydrophones and a geophone, which we used during an activity of listening and gathering inspired by Pauline Oliveros’ practice of Deep Listening, and Jennifer Walshe’s practice as taught to Konstantinos some years ago. Specifically, at certain points, we decided to stop, dip the hydrophones in the river or attach the geophone on resonant surfaces, and listen collectively while one participant at the time was wearing headphones and listening to the hydro/geo-soundworld. After these listening activities, the participants were asked to contribute textual descriptions – little ‘proxy stories’ that ‘amplified and sensitised’ the listening experience – which was later included in the ‘linguistic soup’ mentioned earlier.

Additional sounds were later recorded, when participants became more comfortable in our 'confluence’, and started desiring to contribute certain sounds. These included the reading of an excerpt from Maggie Nelson’s Bluets, the improvised singing of The Prelude by William Wordsworth, a piano improvisation, and various secrets offered to curved parts of the architecture (of Loggia in King Charles’ Court) recorded after the ‘whispering gallery effect’. The gathered sounds were later com-posed within a Max patch (graphic programming software), making quite a wet (both because of the watery recordings and the resonant ‘muddy’ mixing) and generative (not in the trendy AI sense) soundworld, diffused through a quadraphonic set-up for a spatial and immersive experience. The interaction is something we are still developing for future iterations, but it worked like this:

In Stay Hydrated, we placed the hydrophone in the water tank and the geophone on an object with direct contact with the floor. We directed the audience to ‘Walk towards the water, bend over the tank, speak, whisper or shout the leaky words, or voice a secret otherwise, back into the water. If you’d like, take a seat, take a moment.’ Both the hydrophone and the geophone were linked to the max patch. While the geophone was only triggering the flow of text, the hydrophone was analysed to affect the soundworld. The audio stream was analysed through a Mel-Frequency Cepstral Coefficients algorithm, and later fed into a machine learning classifier. We trained the classifier so that it recognises whether someone is speaking into the water tank or whether only the composed soundworld is the dominant captured element. The probability of a number of the classes was intuitively mapped to various parameters of the generative soundworld, so that whenever someone spoke something into the water (often an intimate and vulnerable moment), the watery soundworld left was reconsidering its stability and calmness, through unknowable responses (especially since the recognition is never guaranteed to be ‘correct’, and the effects is always to an extent fed back as a cause of further turbulence). Thus, (machine) listening here is not a process of linear extraction, but a reciprocal relation of resonance.

Image:

Our approach to image was an extension of Natalia’s exploration of during the Arts Territory residency in Venice, involving hybrids of photography, film, and 3d scans. Additionally, we engaged with Italo Calvino’s story, ‘Reading a Wave’ from Mr Palomar, to prompt the group into the unknowability, and impossibility of framing water as a whole. After reading the story as a group, we went mudlarking on Thames during low tides. We each had a version of a 3D scanning app (Polycam) linked to a common cloud drive. We tried to scan water in its confluence with other present bodies, or traces of this process. Later, we replicated the process, but the role of Thames was taken by one of the dancers, who was wearing some of Natalia’s sculptures. We found that the process of scanning the movement of the human dancer was not necessarily easier than scanning the waves forming on the banks of Thames. We later gathered all the 3d scans into a database to be com-posed later along with the rest of the assets.

The visuals were then choreographed virtually and projected in space as a multiplicity to be puddled with rather than looked at. We only used three projectors, but managed to multiply the images into worlds through the various reflective and refractive surfaces. In the loggia of King Charles’ Court, outside the main installation room, we installed a little projector that aimed at a semi-dome, where a looping reel was aligned with the semi-circular edge and warped inside the curved surface. This video included mainly clips from Venice, acting like a little hint, or a portal, connecting the installation to different places and times. Inside the main room, there was a larger new looping reel, depicting a virtual choreography of dance, the river, and various physical and virtual objects, mixed and puddled with each other. Another projector, was placed above the water tank, aiming inside it, which was the central engine of visual defamiliarization. It projected a generative virtual black void, with floating 3D scans and the text flow mentioned earlier. It was aimed at the bottom of the cubic tank, at an angle, which meant that the visuals were activated at the bottom, at the sides of the tank, were reflected on the water’s surface and arrived on the opposite wall, etc. The only readymade that we used, a virtual 3D plastic duck (stock asset in Max), emphasised the irony of the watery container: the glass object was there to hold inside it all that is placed within it, but everything existed only within its glass walls, again, in the depth of the surface.

Code:

The code enabled the confluence of all the digital assets and the installation space. Each dimension – the text, the visuals, the sound, the architecture – followed its own flow, but also leaked into each other through various transversal connections. For example, the semi-dome projections was coupled with the instruction papers, guiding the audience in the main installation room. There, their steps and other loud low frequencies in the ground activated the flow of wavewriting in the projections, the wavewriting in the projections informed the vocalisations of the participants within the water tank, the vocalisation affected the overall soundworld, which often invoked turbulence or at least affected the movement of the audience, which in turn affected the wavewriting, and so on, coupling all the bodies involved into a kind of shared resonance in confluence.